Psychological Review
1993, Vol. 100. No. 3, 363-406
Copyright 1993 by the American Psychological Association, Inc.
OO33-295X/93/S3.OO
The Role of Deliberate Practice in the Acquisition of Expert Performance
K. Anders Ericsson, Ralf Th. Krampe, and Clemens Tesch-Romer
The theoretical framework presented in this article explains expert performance as the end result of
individuals' prolonged efforts to improve performance while negotiating motivational and external
constraints. In most domains of expertise, individuals begin in their childhood a regimen of
effortful activities (deliberate practice) designed to optimize improvement. Individual differences,
even among elite performers, are closely related to assessed amounts of deliberate practice. Many
characteristics once believed to reflect innate talent are actually the result of intense practice
extended for a minimum of 10 years. Analysis of expert performance provides unique evidence on
the potential and limits of extreme environmental adaptation and learning.
Our civilization has always recognized exceptional individ-
uals, whose performance in sports, the arts, and science is
vastly superior to that of the rest of the population. Specula-
tions on the causes of these individuals' extraordinary abilities
and performance are as old as the first records of their achieve-
ments. Early accounts commonly attribute these individuals'
outstanding performance to divine intervention, such as the
influence of the stars or organs in their bodies, or to special
gifts (Murray, 1989). As science progressed, these explanations
became less acceptable. Contemporary accounts assert that the
characteristics responsible for exceptional performance are in-
nate and are genetically transmitted.
The simplicity of these accounts is attractive, but more is
needed. A truly scientific account of exceptional performance
must completely describe both the development leading to ex-
ceptional performance and the genetic and acquired character-
istics that mediate it. This account must specify the critical
differences between exceptional and ordinary performers. It
must also show that any postulated genetic differences can be
hereditary and are plausible from an evolutionary perspective.
Theoreticians in behavioral genetics (Plomin, DeFries, &
McClearn, 1990) now argue that this is a very challenging task
K. Anders Ericsson, Institute of Cognitive Science, University of
Colorado at Boulder; Ralf Th. Krampe and Clemens Tesch-Romer,
Max Planck Institute for Human Development and Education, Berlin,
Federal Republic of Germany.
The empirical research for this article was conducted at the Max
Planck Institute for Human Development in Berlin. Research support
by the Max Planck Society and support and encouragement from Paul
Baltes are gratefully acknowledged.
We thank Peter Usinger and Stefanie Heizmann for their help in the
data collection and Catherine Ashworth, Gregory Carey, Robert
Crutcher, Janet Grassia, Reid Hastie, Stefanie Heizmann, Charles
Judd, Ronald Kellogg, Robert Levin, Clayton Lewis, William Oliver,
Peter Poison, Robert Rehder, Kurt Schlesinger, Vivian Schneider, and
James Wilson for their valuable comments on earlier drafts of this
article. Helpful suggestions and valuable criticism by Richard Shiffrin
on previously submitted versions of this article are gratefully acknowl-
edged.
Correspondence concerning this article should be sent to K. Anders
Ericsson, who is now at the Department of Psychology R-54, Florida
State University, Tallahassee, Florida 32306-1051.
because observed behavior is the result of interactions between
environmental factors and genes during the extended period of
development. Therefore, to better understand expert and ex-
ceptional performance, we must require that the account spec-
ify the different environmental factors that could selectively
promote and facilitate the achievement of such performance. In
addition, recent research on expert performance and expertise
(Chi, Glaser, & Farr, 1988; Ericsson & Smith, 1991a) has shown
that important characteristics of experts' superior performance
are acquired through experience and that the effect of practice
on performance is larger than earlier believed possible. For this
reason, an account of exceptional performance must specify
the environmental circumstances, such as the duration and
structure of activities, and necessary minimal biological attrib-
utes that lead to the acquisition of such characteristics and a
corresponding level of performance.
An account that explains how a majority of individuals can
attain a given level of expert performance might seem inher-
ently unable to explain the exceptional performance of only a
small number of individuals. However, if such an empirical
account could be empirically supported, then the extreme
characteristics of experts could be viewed as having been ac-
quired through learning and adaptation, and studies of expert
performance could provide unique insights into the possibili-
ties and limits of change in cognitive capacities and bodily
functions. In this article we propose a theoretical framework
that explains expert performance in terms of acquired charac-
teristics resulting from extended deliberate practice and that
limits the role of innate (inherited) characteristics to general
levels of activity and emotionality. We provide empirical sup-
port from two new studies and from already published evi-
dence on expert performance in many different domains.
Brief Historical Background
Sir Francis Galton was the first scientist to investigate the
possibility that excellence in diverse fields and domains has a
common set of causes. He found that eminent individuals in
the British Isles were more likely to have close relatives who
were also eminent—although not necessarily in the same do-
main—than to have distant relatives who were eminent. He
concluded that eminence, that is, exceptional performance in a
364
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
field, must be transmitted from parents to their offspring. Gal-
ton (1869/1979) argued that eminence was a virtually inevita-
ble consequence of inherited "natural ability," which was the
conjunction of three types of elements:
By natural ability, I mean those qualities of intellect and disposi-
tion, which urge and qualify a man to perform acts that lead to
reputation. I do not mean capacity without zeal, nor zeal without
capacity, nor even a combination of both of them, without an
adequate power of doing a great deal of very laborious work. (p.
37)
If a man is gifted with vast intellectual ability, eagerness to
work, and power of working, I cannot comprehend how such a
man should be repressed, (p. 39)
Galton readily acknowledged the importance of physiological
exercises for improvement in motor performance and drew a
direct analogy to improvement of mental powers through study-
ing and education. In his view, hereditary factors determine the
limit of the attainable performance for a given individual:
So long as he is a novice, he perhaps flatters himself there is hardly
an assignable limit to the education of his muscles; but the daily
gain is soon discovered to diminish, and at last it vanishes alto-
gether. His maximum performance becomes a rigidly determin-
ate quantity. (Galton, 1869/1979, p. 15)
Even a hundred years later, Galton's conceptualization of emi-
nent performance as reflecting a higher level of ultimate perfor-
mance determined primarily by innate capacities (talent) is still
the modal view among people outside genetics and behavioral
genetics. Genetic influences are still incorrectly viewed as de-
terministic factors that lead to unmodifiable consequences de-
termining the structure of the human body and its nervous
system (Plomin, 1991). Galton's recognition of the interaction
between environmental and genetic factors is clearly shown in
his tri-part definition of natural ability as innate capacity, zeal,
and power to do very laborious work. The last two factors are
also likely to have a genetic component as we argue later in this
article. Nonetheless, the study of eminent performance subse-
quent to Galton has given far less emphasis to zeal and power to
do very laborious work and has focused primarily on genetic
influences on structure and capacities. Everyone agrees that
the shared characteristics of the human body and its nervous
system are due to shared genes. Similarly, the successful identi-
fication of genetic factors influencing individual differences in
height and other physical characteristics has inspired re-
searchers to search for genetic mechanisms regulating individ-
ual differences in mental capacities. Hence the focus of re-
search on talent has been on finding similar basic structural
differences in the nervous system that might mediate stable
differences in expert performance.
Natural Abilities and Other Stable Characteristics
If genetic factors rigidly determine maximal performance, it
is reasonable to assume that these genetic factors cannot be
influenced by practice and training and hence remain stable
across time. Early genetic research showed that many physical
and anatomical attributes, such as height and facial features, are
largely determined by hereditary factors. In many sports the
height of elite athletes is systematically different from that of
the normal population. Greater height is an obvious advantage
in basketball, high jumping, and most sports emphasizing
strength. Shorter height is an advantage in gymnastics. Differ-
ences in height were found to discriminate well among male
athletes of different events at the Olympic games in Montreal,
although the average height of all athletes did not differ from
that of a control group of students (Carter, Ross, Aubrey, Heb-
belinck, & Borms, 1982). Elite athletes also differ in the size of
their muscles, such as arm girth, and in the amount of fat mea-
sured by skin folds. Endurance athletes have a much higher
aerobic ability, larger hearts, more capillaries supplying blood
to muscles, and a higher percentage of slow-twitch muscle
fibers (Ericsson, 1990). Until quite recently researchers com-
monly believed that percentages of muscle fiber types and aer-
obic power "are more than 90% determined by heredity for
males and females" (Brown & Mahoney, 1984, p. 609). Some
researchers have therefore reasoned by analogy that basic gen-
eral characteristics of the nervous system, such as speed of
neural transmission and memory capacities, have a genetic ori-
gin and cannot be changed through training and practice.
Early efforts to find stable individual differences in neural
transmission speed with simple response time (RT) and other
basic capacities were remarkably unsuccessful (Guilford, 1967).
Binet (Varon, 1935) started out using tests of basic perceptual
and cognitive capacities to measure IQ, but found large practice
effects, which were later documented by Gibson (1969). Binet
eventually developed successful IQ tests derived from tests
measuring comprehension, knowledge, and acquired skills. Be-
cause IQ reflects both environmental and genetic factors, re-
cent research has challenged its interpretation and relation to
successful performance outside the school environment (Ceci,
1990; Howe, 1990). The relation of IQ to exceptional perfor-
mance is rather weak in many domains, including music
(Shuter-Dyson, 1982) and chess (Doll & Mayr, 1987). For scien-
tists, engineers, and medical doctors that complete the required
education and training, the correlations between ability mea-
sures and occupational success are only around 0.2, accounting
for only 4% of the variance (Baird, 1985). More generally, pre-
diction of occupational success from psychometric tests has not
been very successful (Tyler, 1965). In a review of more than one
hundred studies, Ghiselli (1966) found the average correlation
between success-on-the-job measuring and aptitude-test scores
to be 0.19. Aptitude tests can predict performance immediately
after training with an average correlation of 0.3, but the correla-
tion between performance after training and final performance
on the job is only about 0.2 (Ghiselli, 1966). Reviews of subse-
quent research (Baird, 1985; Linn, 1982) have reported very
similar correlation estimates. When corrections were made for
the restriction of range of these samples and for unreliability of
performance measures, Hunter and Hunter (1984) found that
only cognitive ability emerged as a useful predictor with an
average adjusted correlation of 0.5 with early job performance.
However, a recent review (Hulin, Henry, & Noon, 1990) has
shown that with increased experience on the job the predictive
validities of ability tests for performance decrease over time by
an average correlation of 0.6 (after corrections for restrictions of
range and unreliability of performance measures). This implies
that ability tests can predict early performance on a job,
whereas final performance is poorly predicted. Even for a well-
THE ROLE OF DELIBERATE PRACTICE
365
defined skill, such as typing, with relatively unselect groups of
subjects, numerous efforts to predict the attained performance
from pretraining aptitude tests have failed (Clem, 1955). Strik-
ing differences between eminent individuals (experts) and less
accomplished individuals are found, not surprisingly, when
their current performance in the field of expertise is compared
(Ericsson & Smith, 1991 b); experts are faster and more accurate
than less accomplished individuals. However, experts' superior
speed in their domain of expertise does not transfer to general
tests of speed, such as simple RT, or to general tests of percep-
tion (Starkes, 1987; Starkes & Deakin, 1984). Similarly, experts'
memory for representative stimuli from their domain is vastly
superior to that of lesser experts, especially for briefly pre-
sented stimuli. But when tested on randomly rearranged ver-
sions of representative stimuli from their domain presented
with short exposures or on materials outside their domain, the
memory of experts is no better than that of ordinary individ-
uals.
The domain-specific nature of experts' superior perfor-
mance implies that acquired knowledge and skill are important
to attainment of expert performance. We can cite only two abili-
ties that investigators have argued directly reflect genetic fac-
tors. Some successful musicians can recognize a musical note in
isolation by its pitch (perfect pitch). Championship-level ty-
pists can tap their fingers faster than normal (Book, 1924;
Keele & Hawkins, 1982). Although we claim that genetic fac-
tors have little direct impact on ultimate adult performance, a
plausible role for hereditary factors is in the developmental his-
tory of an individual. Superior performance by very young
children without prior instruction may suggest exceptional
promise, leading to the early onset of training. This in turn
leads to a consistently greater accumulation of practice (and
hence, by our framework, performance) relative to later-start-
ing individuals. In the General Discussion section we consider
this potential indirect role of innate talent at length, concluding
instead that unique environmental conditions and parental
support, rather than talent, may be the important factors deter-
mining the initial onset of training and ultimate performance.
In summary, the search for stable heritable characteristics
that could predict or at least account for the superior perfor-
mance of eminent individuals has been surprisingly unsuccess-
ful. The best evidence for the effect of heritable characteristics
comes from several types of sports, for which anatomical char-
acteristics such as height systematically differ for elite per-
formers compared with the average population. The belief that
the striking differences between expert performers and less
accomplished performers reflect innate abilities (talent) is so
strong that the failure to identify the specific talents necessary
for expert performance in a given domain is viewed, at most, as
a temporary problem until the relevant talents are discovered.
The conviction in the importance of talent appears to be based
on the insufficiency of alternative hypotheses to explain the
exceptional nature of expert performance. If one agrees with
Galton's plausible claim that the improvements resulting from
experience and practice occur during limited time until a stable
maximal level of performance is attained, the factors limiting
further improvement must be fixed and unmodifiable by envi-
ronmental factors. The most likely source of such unmodifi-
able factors is genetic. However, this argument is only valid if
the associated assumptions can be verified empirically.
In the following two sections we examine the assumption
that with extensive experience in a domain a maximal level of
performance is automatically reached and that the period of
improvement has a relatively limited duration, especially for
talented individuals.
Does Practice and Experience Inevitably Lead to
Maximal Performance?
The view that merely engaging in a sufficient amount of
practice, regardless of the structure of that practice, leads to
maximal performance has a long and contested history. In their
classic studies of Morse Code operators, Bryan and Harter
(1897, 1899) identified plateaus in skill acquisition, when for
long periods subjects seemed unable to attain further improve-
ments. However, with extended efforts, subjects could restruc-
ture their skill to overcome plateaus. Keller (1958) later showed
that these plateaus in Morse Code reception were not an inevita-
ble characteristic of skill acquisition, but could be avoided by
different and better training methods. Nonetheless, Bryan and
Harter (1897, 1899) had clearly shown that with mere repeti-
tion, improvement of performance was often arrested at less
than maximal levels, and further improvement required effort-
ful reorganization of the skill. Even very experienced Morse
Code operators could be encouraged to dramatically increase
their performance through deliberate efforts when further im-
provements were required for promotions and external rewards
(Bryan & Harter, 1897).
More generally, Thorndike (1921) observed that adults per-
form at a level far from their maximal level even for tasks they
frequently carry out. For instance, adults tend to write more
slowly and illegibly than they are capable of doing. Likewise,
adults (including clerks with many years of frequent daily expe-
rience) add numbers far more slowly than they can when they
are doing their best. Thorndike (1921, p. 178) accounts for these
curious observations with the following comment: "It is that we
have too many other improvements to make, or do not know
how to direct our practice, or do not really care enough about
improving, or some mixture of these three conditions." In sup-
port of this claim, he reported several laboratory studies and a
study of experienced typesetters by Aschaffenburg (1896),
which showed gradual improvements of up to 25% as a result of
continued testing. Kitson (as described in Book & Norvell,
1922) found that during a 20-week period, typesetters with
around 10 years of experience gradually improved their job
performance between 58% and 97% in response to a bonus
system rewarding higher performance. Dvorak, Merrick, Dea-
ley, and Ford (1936) reported substantial improvements in expe-
rienced typists as a result of deliberate efforts.
Because performance in sports, especially, has been mea-
sured under standardized conditions, and the best perfor-
mance has been recorded at world, national, district, and club
levels, it can be clearly demonstrated that performance has con-
tinually improved during this entire century. Schulz and Cur-
now (1988) found that throughout the history of the Olympic
Games, the best performance for all events has improved—in
some cases by more than 50%. It is generally recognized that
366
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
some of these improvements are due to equipment and rule
changes, but improvements are great even in events with minor
changes, such as running and swimming. Increases in duration,
intensity, and structure of training appear to play a major role.
The fastest time for the marathon in the 1896 Olympic Games
was just a minute faster than the required entry time in large
marathon races such as the Boston Marathon (Ericsson, 1990).
The fastest rate of typing in the World Championship in typing
increased from 82 words per minute in 1904 to 147 words per
minute in 1923—an improvement of 80% (Book, 1925a). Even
in music there is evidence for improved skill. When Tchai-
kovsky asked two of the greatest violinists of his day to play his
violin concerto, they refused, deeming the score unplayable
(Platt, 1966). Today, elite violinists consider this concerto part
of the standard repertory. The improvement in music training
is so great that according to Roth (1982) the violin virtuoso
Paganini "would indeed cut a sorry figure if placed upon the
modern concert stage" (p. 23).
In virtually all domains, insights and knowledge are steadily
accumulating and the criteria for eminent as well as expert
performance undergo continuous change. To reach the status of
an expert in a domain it is sufficient to master the existing
knowledge and techniques. To make an eminent achievement
one must first achieve the level of an expert and then in addi-
tion surpass the achievements of already recognized eminent
people and make innovative contributions to the domain. In
sum, the belief that a sufficient amount of experience or prac-
tice leads to maximal performance appears incorrect.
Preparation Time Required for Attainment of
Exceptional Performance
There is a relatively widespread conception that if individuals
are innately talented, they can easily and rapidly achieve an
exceptional level of performance once they have acquired basic
skills and knowledge. Biographical material disproves this no-
tion. In their classic study of expertise in chess, Simon and
Chase (1973) observed that nobody had attained the level of an
international chess master (grandmaster) "with less than about
a decade's intense preparation with the game" (p. 402). Simon
and Chase estimated that the amount of knowledge a chess
master has acquired is comparable in size to the vocabulary of
an adult native speaker of English. It takes normal individuals
approximately a decade to acquire this vocabulary. Similarly,
Krogius (1976) showed that the time between chess players'
first learning the rules of chess and attaining international
chess master status was 11.7 years for those who learned chess
rules late (after age 11) and even longer for those who started
early, that is, 16.5 years. If only well-established domains with a
large number of active individuals are considered we know of
only a small number of exceptions to the general rule that indi-
viduals require 10 or more years of preparation to attain interna-
tional-level performance. The exceptions in this century, such
as the famous chess players, Bobby Fischer and Salo Flohr, were
only a year shy of the prerequisite 10 years of preparation (Kro-
gius, 1976).
J. R. Hayes (1981) confirmed that 10 years' experience is
necessary in another domain, musical composition. He calcu-
lated an average of about 20 years from the time individuals
started to study music until they first composed an outstanding
piece of music. According to Hayes, this long preparation pe-
riod is necessary because "the composer must know the
timbres of the various instruments and the sound, look, and
feel of chords and key structures" (p. 209). Most important,
Hayes showed that the 10 or more years of necessary experience
was not an artifact. Because musicians start very early, insuffi-
cient development may restrict their ability to compose before
attaining adulthood. Those who started at ages younger than 6
years did not write their first eminent composition until 16.5
years later; those who started between ages 6 and 9 and older
than 10 years of age required 22 and 21.5 years, respectively, to
compose their first distinguished work. Simon and Chase's
(1973) "10-year rule" is supported by data from a wide range of
domains: music (Sosniak, 1985), mathematics (Gustin, 1985),
tennis (Monsaas, 1985), swimming (Kalinowski, 1985), and
long-distance running (Wallingford, 1975).
Long periods of necessary preparation can also be inferred
for writers and scientists, although the starting point of their
careers is more difficult to determine. Scientists have reported
that they made a career decision during their middle or late
teens, whereas they most often published a truly major contri-
bution one or two decades later (Lehmann, 1953). Raskin
(1936), who analyzed the 120 most important scientists and 123
most famous poets and authors in the 19th century, found that
the average age at which scientists published their first work
was 25.2; poets and authors published their first work at the
average age of 24.2. Moreover, many years of preparation pre-
ceded first publication. The average ages at which the same
individuals produced their greatest work were 35.4 for scien-
tists and 34.3 for poets and authors. That is, on average, more
than 10 years elapsed between these scientists' and authors'
first work and their best work. In many other domains, the
highest level of expert performance is displayed by individuals
with more than 10 years of experience: evaluation of livestock
(Phelps&Shanteau, 1978), diagnosis of X-rays (Lesgold, 1984),
and medical diagnosis (Patel & Groen, 1991). This evidence is
consistent with Galton's (1869/1979) claim that motivation and
perseverance are necessary for attainment of eminent perfor-
mance.
Our review has also shown that the maximal level of perfor-
mance for individuals in a given domain is not attained auto-
matically as function of extended experience, but the level of
performance can be increased even by highly experienced indi-
viduals as a result of deliberate efforts to improve. Hence, stable
levels of performance after extended experience are not rigidly
limited by unmodifiable, possibly innate, factors, but can be
further increased by deliberate efforts. We have shown that ex-
pert performance is acquired slowly over a very long time as a
result of practice and that the highest levels of performance and
achievement appear to require at least around 10 years of in-
tense prior preparation. However the relation between ac-
quired performance and the amount of practice and experience
was found to be weak to moderate in the earlier review. We
propose that the reason for this comparatively weak relation is
that the current definition of practice is vague. If we are to
improve our understanding of the environmental influences
mediated through participation in different activities, we must
analyze the types of activities commonly called practice.
THE ROLE OF DELIBERATE PRACTICE
367
The Role of Deliberate Practice
In this section we characterize deliberate practice—those ac-
tivities that have been found most effective in improving perfor-
mance. We then contrast deliberate practice with activities that
tend to occur more frequently in various domains. Finally, we
propose a theoretical framework that explains how expert per-
formance can be attained through deliberate practice.
Characteristics of Deliberate Practice
The basic skills required for living in a culture are acquired
by virtually all children as part of normal social interaction
with a minimum of instruction. In contrast, the skills of read-
ing, writing, and arithmetic have been explicitly taught in
schools by teachers with assigned activities of, for example,
copying of presented material, for more than 3 thousand years
(Eby & Arrowood, 1940). We want to distinguish activities in-
vented with the primary purpose of attaining and improving
skills from other types of everyday activities, in which learning
may be an indirect result. On the basis of several thousand
years of education, along with more recent laboratory research
on learning and skill acquisition, a number of conditions for
optimal learning and improvement of performance have been
uncovered (Bower & Hilgard, 1981; Gagne, 1970). The most
cited condition concerns the subjects' motivation to attend to
the task and exert effort to improve their performance. In addi-
tion, the design of the task should take into account the preex-
isting knowledge of the learners so that the task can be correctly
understood after a brief period of instruction. The subjects
should receive immediate informative feedback and knowledge
of results of their performance. The subjects should repeatedly
perform the same or similar tasks.
When these conditions are met, practice improves accuracy
and speed of performance on cognitive, perceptual, and motor
tasks (Fitts & Posner, 1967; Gibson, 1969; Welford, 1968). Tasks
used in laboratory studies of learning that are designed to focus
on the accuracy of performance clearly display the relevant cues
and the relevant feedback. Studies focusing on speed of perfor-
mance tend to use easy tasks, where highly accurate perfor-
mance is rapidly attained, and subjects are instructed to in-
crease the speed of performance while maintaining the high
level of accuracy. Under these conditions subjects' performance
improves monotonically as a function of the amount of practice
according to the power law (J. R. Anderson, 1982; Newell &
Rosenbloom, 1981). In the absence of adequate feedback, effi-
cient learning is impossible and improvement only minimal
even for highly motivated subjects. Hence mere repetition of an
activity will not automatically lead to improvement in, espe-
cially, accuracy of performance (Trowbridge & Cason, 1932).
When laboratory training is extended over longer time pe-
riods, studies show that providing a motivated individual with
repeated exposure to a task does not ensure that the highest
levels of performance will be attained. Assessment of subjects'
methods shows that inadequate strategies often account for the
lack of improvement. For example, in their study on the effects
of practice on digit span, Chase and Ericsson (1981) found a
subject who kept rehearsing the digits whose performance
showed only minimal improvement. In contrast, all subjects
who used preexisting knowledge to encode the presented digits
improved dramatically. One subject who discovered how to use
efficient retrieval structures increased his performance by over
1000%. Recent reviews of exceptional memory performance
(Ericsson, 1985,1988) show that a small set of general methods
underlie such performance. After being instructed to use ade-
quate strategies, subjects have attained exceptional levels of
memory performance after extended practice (Baltes & Kliegl,
1992; Kliegl, Smith, & Baltes, 1989; 1990).
Early investigators of extended skill acquisition in typing
(Book, 1925b; Dvorak et al., 1936) and other perceptual-motor
skills (Kao, 1937) carefully monitored improvements in perfor-
mance and collected verbal reports on subjects' cognitive pro-
cesses. These studies revealed subjects' active search for meth-
ods to improve performance and found that changes in meth-
ods could often be related to clear improvements. Other studies
(Chase & Ericsson, 1981; VanLehn, 1991) have also shown that
subjects actively try out different methods and refine methods
in response to errors and violated expectations. The critical
importance of a correct method or strategy has also been dem-
onstrated in date calculation (Addis & O. A. Parsons, as de-
scribed in Ericsson & Faivre, 1988), mental multiplication
(Chase & Ericsson, 1982; Staszewski, 1988), absolute judgment
of colors and pitches (for a review see Ericsson & Faivre, 1988),
motor skills (Norman, 1976), and methods of work (R. H. Sea-
shore, 1939).
The inability of some subjects to discover new methods has
sometimes been interpreted as evidence for basic cognitive or
perceptual deficits, especially for performance of seemingly
simple tasks. However, specific instruction or the generation of
new methods can eventually enhance improvement tempo-
rarily arrested at suboptimal levels. As the complexity of a de-
sired skill increases beyond the simple structure of most labora-
tory tasks, the logically possible methods to correctly and incor-
rectly perform the task by subjects increase as well. To assure
effective learning, subjects ideally should be given explicit in-
structions about the best method and be supervised by a
teacher to allow individualized diagnosis of errors, informative
feedback, and remedial part training. The instructor has to
organize the sequence of appropriate training tasks and moni-
tor improvement to decide when transitions to more complex
and challenging tasks are appropriate. Although it is possible to
generate curricula and use group instruction, it is generally
recognized that individualized supervision by a teacher is supe-
rior. Research in education reviewed by Bloom (1984) shows
that when students are randomly assigned to instruction by a
tutor or to conventional teaching, tutoring yields better perfor-
mance by two standard deviations—the average tutored stu-
dent performed at the 98th percentile of students taught with
the conventional method. Interestingly, the correlation between
prior achievement and achievement on the current course was
reduced and corresponded to only about 6% of the variance for
the tutored subjects as compared with around 36% for students
taught with conventional methods. More generally, improved
instruction appears to benefit subjects with lower cognitive ability
more than high-ability subjects thus lowering the earlier
discussed correlation between cognitive ability and early perfor-
mance seen under standard training conditions.
Most contemporary domains of expertise have evolved over
368
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
centuries from activities originally centered around playful in-
teraction with learning through active participation. As the lev-
els of performance in the domain increased in skill and com-
plexity, methods to explicitly instruct and train individuals
were developed. In all major domains there has been a steady
accumulation of knowledge about the best methods to attain a
high level of performance and the associated practice activities
leading to this performance. Full-time teachers and coaches are
available for hire and supervise the personalized training of
individuals at different levels of performance starting with be-
ginners. Throughout development toward expert performance,
the teachers and coaches instruct the individuals to engage in
practice activities that maximize improvement. Given the cost
of individualized instruction, the teacher designs practice activ-
ities that the individual can engage in between meetings with
the teacher. We call these practice activities deliberate practice
and distinguish them from other activities, such as playful in-
teraction, paid work, and observation of others, that individ-
uals can pursue in the domain.
Comparison of Deliberate Practice to Other Types of
Domain-Related Activities
Consider three general types of activities, namely, work, play,
and deliberate practice. Work includes public performance,
competitions, services rendered for pay, and other activities di-
rectly motivated by external rewards. Play includes activities
that have no explicit goal and that are inherently enjoyable.
Deliberate practice includes activities that have been specially
designed to improve the current level of performance. The
goals, costs, and rewards of these three types of activities differ,
as does the frequency with which individuals pursue them.
Public performance and competitions are constrained in
time; these activities as well as rendering a service for pay re-
quire that individuals give their best performance at a given
time. The distinction between work and training (deliberate
practice) is generally recognized. Individuals given a new job
are often given some period of apprenticeship or supervised
activity during which they are supposed to acquire an accept-
able level of reliable performance. Thereafter individuals are
expected to give their best performance in work activities and
hence individuals rely on previously well-entrenched methods
rather than exploring alternative methods with unknown reli-
ability. The costs of mistakes or failures to meet deadlines are
generally great, which discourages learning and acquisition of
new and possibly better methods during the time of work. For
example, highly experienced users of computer software appli-
cations are found to use a small set of commands, thus avoiding
the learning of a larger set of more efficient commands (see
Ashworth, 1992, for a review). Although work activities offer
some opportunities for learning, they are far from optimal. In
contrast, deliberate practice would allow for repeated experi-
ences in which the individual can attend to the critical aspects
of the situation and incrementally improve her or his perfor-
mance in response to knowledge of results, feedback, or both
from a teacher. Let us briefly illustrate the differences between
work and deliberate practice. During a 3-hr baseball game, a
batter may get only 5-15 pitches (perhaps one or two relevant to
a particular weakness), whereas during optimal practice of the
same duration, a batter working with a dedicated pitcher has
several hundred batting opportunities, where this weakness can
be systematically explored (T. Williams, 1988).
The external rewards of work activities include social recogni-
tion and, most important, money in the form of prizes and pay,
which enables performers to sustain a living. In play and deliber-
ate practice, external rewards are almost completely lacking.
The goal of play is the activity itself, and the inherent enjoyment
of it is evident in children who spontaneously play for extended
periods of time. Recent analyses of inherent enjoyment in
adults reveal an enjoyable state of "flow," in which individuals
are completely immersed in an activity (Csikszentmihalyi,
1990). Similarly, analyses of reported "peak experiences" in
sports reveal an enjoyable state of effortless mastery and execu-
tion of an activity (Ravizza, 1984). This state of diffused atten-
tion is almost antithetical to focused attention required by delib-
erate practice to maximize feedback and information about
corrective action.
In contrast to play, deliberate practice is a highly structured
activity, the explicit goal of which is to improve performance.
Specific tasks are invented to overcome weaknesses, and perfor-
mance is carefully monitored to provide cues for ways to im-
prove it further. We claim that deliberate practice requires ef-
fort and is not inherently enjoyable. Individuals are motivated
to practice because practice improves performance. In addi-
tion, engaging in deliberate practice generates no immediate
monetary rewards and generates costs associated with access to
teachers and training environments. Thus, an understanding of
the long-term consequences of deliberate practice is important.
Theoretical Framework for the Acquisition of Expert
Performance
We now outline a framework within which we can explain
how differential levels of performance are attained as a func-
tion of deliberate practice. Our basic assumption—the "mono-
tonic benefits assumption"—is that the amount of time an indi-
vidual is engaged in deliberate practice activities is monotoni-
cally related to that individual's acquired performance. This
assumption can be tested empirically. It follows from this as-
sumption that individuals should attempt to maximize the
amount of time they spend on deliberate practice to reach ex-
pert performance.
However, maximization of deliberate practice is neither
short-lived nor simple. It extends over a period of at least 10
years and involves optimization within several constraints.
First, deliberate practice requires available time and energy for
the individual as well as access to teachers, training material,
and training facilities (the resource constraint). If the individual
is a child or adolescent, someone in the individual's environ-
ment must be willing to pay for training material and the time
of professional teachers, as well as for transportation to and
from training facilities and competitions.
Second, engagement in deliberate practice is not inherently
motivating. Performers consider it instrumental in achieving
further improvements in performance (the motivational con-
straint). The lack of inherent reward or enjoyment in practice as
distinct from the enjoyment of the result (improvement) is con-
THE ROLE OF DELIBERATE PRACTICE
369
sistent with the fact that individuals in a domain rarely initiate
practice spontaneously.
Finally, deliberate practice is an effortful activity that can be
sustained only for a limited time each day during extended
periods without leading to exhaustion (effort constraint). To
maximize gains from long-term practice, individuals must
avoid exhaustion and must limit practice to an amount from
which they can completely recover on a daily or weekly basis.
Attaining Expert Performance
Considering the cost of pursuing expert-level performance
and the small number of individuals who, out of millions of
children exposed to such domains as sports and music, can
make a living as professionals, it seems remarkable that individ-
uals get started and are encouraged to continue. From many
interviews with international-level performers in several do-
mains, Bloom (1985b) found that these individuals start out as
children by engaging in playful activities in the domain. After
some period of playful and enjoyable experience they reveal
"talent" or promise. At this point parents typically suggest the
start of instruction by a teacher and limited amounts of deliber-
ate practice. The parents support their children in acquiring
regular habits of practice and teach their children about the
instrumental value of deliberate practice by noticing improve-
ments in performance. With increased experience and deliber-
ate practice, individuals' performance in the domain reflects
an inseparable combination of practice and innate talent. We
rely on Bloom's (1985b) characterization of the period of prepa-
ration in three phases, which are illustrated in Figure 1.
The first phase begins with an individual's introduction to
activities in the domain and ends with the start of instruction
and deliberate practice. The second phase consists of an ex-
tended period of preparation and ends with the individual's
commitment to pursue activities in the domain on a full-time
basis. The third phase consists of full-time commitment to im-
proving performance and ends when the individual either can
make a living as a professional performer in the domain or
terminates full-time engagement in the activity. During all
three phases the individual requires support from external
sources, such as parents, teachers, and educational institutions.
This framework needs to be extended with a fourth phase to
accommodate eminent performance. During this fourth phase
the individuals go beyond the knowledge of their teachers to
make a unique innovative contribution to their domain.
To be complete, our theoretical framework must show how
individuals negotiate the various constraints on deliberate
practice during that first decade of preparation necessary for
attaining international-level performance. There are several
methodological problems involved in demonstrating the rele-
vant processes. During the decade or two leading to adult ex-
pert performance, many aspects of training and evaluation
change. In the beginning, a child's performance is compared
with that of other children of the same age in the local neigh-
borhood. At the start of participation in competitions, the refer-
ence group consists of other trained individuals of similar ages
from a larger area. Success at these earlier stages may eventually
lead to participation in competitions at a national and interna-
tional level. At increased levels of performance, the practice
activities obviously change and so do the criteria of evaluation.
In the performance of music, children and adolescents are
judged principally on their technical proficiency. Expert adult
performers, however, are judged on their interpretation and
ability to express emotions through music (Sloboda, 1991). The
inability of many child prodigies in music to succeed as adult
musicians (Bamberger, 1986; Barlow, 1952) is often attributed
to difficulties making this transition—possibly resulting from
inappropriate training and instruction during the early and
middle phases of music training. To become outstanding musi-
cians at the international level, individuals have to contribute
unique interpretations of music (Roth, 1982). Similar consider-
ations may explain why mathematical prodigies can fail as
adult mathematicians. The lack of overlap in the performance
of precocious children and adult scientists in mathematics is
even clearer than in music: Superior ability in mental addition
370
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
and multiplication demonstrate efficiency in the mechanics of
mathematics, whereas major adult contributions in mathemat-
ics reflect insights into the structure of mathematical problems
and domains. The criteria for eminent performance goes
beyond expert mastery of available knowledge and skills and
requires an important and innovative contribution to the do-
main. An eminent musician can contribute new techniques and
distinct interpretations of existing music, and eminent chess
players discover new variants of chess openings and advance
the knowledge of chess. In the arts and sciences, eminent
achievements involve contributions of new ideas, theories, and
methods.
In most domains it is impossible to assess retrospectively the
cognitive aspects of the development of precocious exceptional
performers. Precocious painters may be an exception (see J.
Radford's, 1990, review). Pariser (1987) analyzed drawings
completed by Klee, Toulouse-Lautrec, and Picasso until age 20,
and concluded that these three "gifted" individuals encoun-
tered and mastered problems in graphic development in ways
similar to those of the "less-gifted" (p. 53). Their juvenile draw-
ings include "a fair number of awkward, flawed and unexcep-
tional drawings" (Pariser, 1987, p. 65), suggesting that instruc-
tion and practice strongly affected even these three exceptional
artists. The age at which eminent individuals attain their best
performance is much later in their 20s and 30s (Lehmann,
1953). In fact any significant achievements in literature, music
composition, visual arts, and most other domains before age 16
are exceedingly rare (Barlow, 1952). (Judit Polgar and Bobbie
Fischer attained the level of international grandmasters in
chess at age 15; we discuss their developmental history later.)
The methodology we applied in our studies takes these consid-
erations into account as we demonstrate later. At this point we
must further specify the constraints inherent in the attainment
of exceptional performance.
Resource Constraint
International-level performers often receive their first expo-
sure to their domain between the ages of 3 and 8. Obviously,
their parents are responsible for providing this early access.
Parents and guardians, in encouraging the childrens' activity
and monitoring performance, make possible the discovery of
early signs of "talent" and promise. The parents' interest is also
critical in aiding children's transition to deliberate practice and
providing facilities for practice, such as musical instruments for
musicians, tennis courts for tennis players, and ice arenas for
skaters. Bloom (1985a) and his colleagues show that transporta-
tion for young individuals to and from practice, meetings with
the teachers, and competitions can almost completely occupy
parents' free time, and the direct economic costs of sustaining
these activities are substantial. The parents' costs for a national-
level swimmer is estimated by Chambliss (1988) to exceed 5
thousand dollars per year. In many cases, the family is even
willing to move to a location close to the best training facilities
offering year-round opportunities for practice. These extraordi-
nary commitments by parents are probably based on the belief
that their children are somehow special and particularly likely
to succeed. Bloom (1985b) found that there seems to be at least
one central person in a promising child's near environment who
firmly believes, as the child develops, that the child is special,
that is, talented in the domain. This person's belief prevails
even though Bloom (1985b) found no evidence that, during the
early phases, the individual exhibited any clear evidence of pro-
wess. However, Bloom (1985b) found that only one child per
family was considered special. This is perhaps the best empiri-
cal evidence that each family's available resources are limited.
Effort Constraint
The central claim of our framework is that the level of perfor-
mance an individual attains is directly related to the amount of
deliberate practice. Hence, individuals seeking to maximize
their performance within some time period should maximize
the amount of deliberate practice they engage in during that
period. When this time period extends over months and years,
it is clear that maximization of an effortful activity is not simple
and that the traditional research on learning, which is limited
to a few sessions, provides little guidance. In this section, we
review evidence showing that the duration of effective daily
practice that can be sustained for long periods is limited, and
that according to teachers and training instructions, it is neces-
sary to maintain full attention during the entire period of delib-
erate practice. We then discuss some consequences of increas-
ing practice activity beyond its optimal duration and finally
consider evidence that through training the daily duration of
deliberate practice can be slowly increased over extended pe-
riods of time.
The limited duration of practice is the best evidence of the
effort it requires. When individuals, especially children, start
practicing in a given domain, the amount of practice is an hour
or less per day (Bloom, 1985b). Similarly, laboratory studies of
extended practice limit practice to about 1 hr for 3-5 days a
week (e.g., Chase & Ericsson, 1982; Schneider & Shiffrin, 1977;
Seibel, 1963). A number of training studies in real life have
compared the efficiency of practice durations ranging from 1 -8
hr per day. These studies show essentially no benefit from dura-
tions exceeding 4 hr per day and reduced benefits from practice
exceeding 2 hr (Welford, 1968; Woodworth & Schlosberg,
1954). Many studies of the acquisition of typing skill (Baddeley
& Longman, 1978; Dvorak et al.. 1936) and other perceptual-
motor skills (Henshaw & Holman, 1930) indicate that the effec-
tive duration of deliberate practice may be closer to 1 hr per day.
Pirolli and J. R. Anderson (1985) found no increased learning
from doubling the number of training trials per session in their
extended training study. The findings of these studies can be
generalized to situations in which training is extended over long
periods of time such as weeks, months, and years.
The goal of deliberate practice is improved performance, and
detailed analyses of the musicians' activities during practice
sessions in music (Gruson, 1988; Miklaszewski, 1989) reveal
careful monitoring and problem solving by the musicians to
attain the desired improvements. C. E. Seashore (1938/1967),
the pioneering researcher in music psychology, claimed, "Many
a student becomes disgusted with music because he cannot
learn by dull drudgery. The command to rest is fully as impor-
tant as to work in effective learning" (pp. 154-155). Both Auer
(1921), the famous violin teacher, and C. E. Seashore (1938/
1967) recommended that practice periods be limited to less
THE ROLE OF DELIBERATE PRACTICE
371
than 1 hr with ample rest in between. A necessary precondition
for practice, according to Auer (1921), is that the individual be
fully attentive to his playing so that he or she will notice areas of
potential improvement and avoid errors. Auer (1921) believes
that practice without such concentration is even detrimental to
improvement of performance. On the basis of an extended
study of Olympic swimmers, Chambliss (1988, 1989) argued
that the secret of attaining excellence is to always maintain
close attention to every detail of performance "each one done
correctly, time and again, until excellence in every detail be-
comes a firmly ingrained habit" (1989, p. 85).
Deliberate practice aimed at improving strength and endur-
ance in sports clearly shows the importance of near maximal
effort during practice and the resulting fatigue. Physical activity
and exercise produce no benefit unless they are sufficiently
intense. Untrained adults must attain a minimum heart rate of
around 140 beats per minute or 70% of their maximal heart rate
for an extended time at least three times a week to see improve-
ments (Lamb, 1984). However, elite athletes train at much
higher intensities to improve their performance. Athletes train
to maximize their performance in a specific event. In endur-
ance events, such as marathon running, most of the training
consists of running at the highest speed an athlete can maintain
for extended periods. Improvements resulting from training ap-
pear to be more a function of intensity (as close to maximum as
possible) than of the total distance covered (Maughan, 1990). In
sprint events, where runners expend maximal effort for a short
time, strength training is essential (P. F. Radford, 1990). Near
maximal efforts with a 3-s duration produce the most efficient
results for strength training (Klausen, 1990). Obviously such
near-maximal training can be sustained only for limited pe-
riods even if these periods are interspersed with periods of rest.
Other objective indicators of the intensity of athletic training
include measurements of metabolic rate during the activity
(MacLaren, 1990). Athletes need to consume many more calo-
ries than do normal adults simply to sustain their regular train-
ing program (Maughan, 1990). Costill et al. (1988) found that
some swimmers experienced chronic muscular fatigue because
their intake of calories was insufficient to accommodate a re-
cent increase in training activity. The exhausting effects of regu-
lar training are also evidenced by the standard practice of re-
ducing the training level several days before a competition
(Maughan, 1990; P. E Radford, 1990).
Under the assumption that practice draws on limited physi-
cal and mental resources, one would expect that the level of
practice an individual can sustain for long periods of time is
limited by the individual's ability to recover and thereby main-
tain a steady state from day to day. After the individual has
slowly adapted to a constant level of practice, increases ought to
be possible. In contrast, if an individual cannot recover each
day from a given level of practice, sustaining that level will lead
to exhaustion and mental fatigue. The risk of physical injury
and chronic maladaptation will increase "runner's knee," shin
splints, and Achilles tendonitis for athletes (Subotnick, 1977)
and sores, tendonitis, and muscle spasms for musicians (Cal-
dron et al., 1986). Inability to recover from the stress of training,
which is viewed as necessary for improvement in sports, can
lead to "staleness," "overtraining," and eventually "burnout."
These states are characterized not only by physical fatigue and
soreness but also by motivational problems such as lack of en-
thusiasm and even unwillingness to continue with a sport
(Silva, 1990). The only known effective treatment for these con-
ditions "consists of rest, and in some cases, complete abstention
from training and sporting activities may be necessary" (Hack-
ney, Pearman, & Novack, 1990, p. 30).
Early in this century, considerable research was directed to-
ward the subjective experience of mental fatigue and its conse-
quences for performance. On the one hand, efforts to demon-
strate decline in performance, even after consecutive days of
mental multiplication for 12 hr per day, have been remarkably
unsuccessful (Arai, 1912; Huxtable, White, & McCartor, 1946).
On the other hand, the subjective feelings of discomfort and
aversion often become so strong that continuing these experi-
ments beyond 4 days would seem very difficult, if not impossi-
ble. The best data on sustained intellectual activity comes from
financially independent authors. While completing a novel fa-
mous authors tend to write only for 4 hr during the morning,
leaving the rest of the day for rest and recuperation (Cowley,
1959; Plimpton, 1977). Hence successful authors, who can con-
trol their work habits and are motivated to optimize their pro-
ductivity, limit their most important intellectual activity to a
fixed daily amount when working on projects requiring long
periods of time to complete.
When individuals start with deliberate practice in a domain,
the initial duration of weekly practice is limited (Bloom,
1985b). Given that most future international-level performers
start at early ages, these brief durations are consistent with the
short duration (10-20 min per session) of long-term training
programs with children (see Howe, 1990, for a review). Consis-
tent with the idea of slow adaptation to the demands of ex-
tended practice, individuals beginning to practice are encour-
aged to adopt a regular weekly schedule with practice periods of
relatively fixed duration (Bloom, 1985b). After extended time
with an acceptable practice level, individuals adapt their bodies
and lives and can slowly and gradually increase the level of
practice. Too rapid increases in the intensity of practice lead to
"overuse and overtraining," which occur frequently in sports
(Hackney et al, 1990; Silva, 1990) and even in music (Fry, 1986;
Newmark & Lederman, 1987). Bailey and Martin (1988) report
many instances of successful 9- to 11-year-old children increas-
ing their training to very high levels, only to experience motiva-
tional burnout and quit the domain altogether.
In summary, disregard of the effort constraint on deliberate
practice leads to injury and even failure. In the short term,
optimal deliberate practice maintains equilibrium between ef-
fort and recovery. In the long term, it negotiates the effort con-
straint by slow, regular increases in amounts of practice that
allow for adaptation to increased demands.
Motivational Constraint
A premise of our theoretical framework is that deliberate
practice is not inherently enjoyable and that individuals are
motivated to engage in it by its instrumental value in improving
performance. Hence, interested individuals need to be engag-
ing in the activity and motivated to improve performance be-
fore they begin deliberate practice. Bloom (1985b) found evi-
dence supporting this implication. His interviews with interna-
372
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
tional-level performers showed that parents typically initiated
deliberate practice after allowing their children several months
of playful engagement in the domain and after noticing that
their children expressed interest and showed signs of promise.
The social reactions of parents and other individuals in the
immediate environment must be very important in establish-
ing this original motivation.
At the start of deliberate practice, parents help their child
keep a regular daily practice schedule and point out the instru-
mental value of practice for improved performance (Bloom,
1985b). With increased experience and the aid of teachers and
coaches, the developing individual is able to internalize meth-
ods for assessing improvement and can thus concurrently mon-
itor the effects of practice. As individuals get more involved in
the activities of a domain, competitions and public perfor-
mances provide short-term goals for specific improvements. At
this point the motivation to practice becomes so closely con-
nected to the goal of becoming an expert performer and so
integrated with the individual's daily life that motivation to
practice, per se, cannot be easily assessed.
Certain naturally occurring events and changes illuminate
the relation between practice and performance. Activities in
many domains, especially sports, are seasonal because most
scheduled competitions occur during a single season of the
year. If individuals enjoyed deliberate practice, they ought to
practice at a uniformly high level all year. Instead, athletes train
much harder during the preseason period and during the sea-
son itself; during the off season they often reduce the level of
training dramatically (Reilly, 1990a; Reilly & Secher, 1990).
Many individuals who have practiced for a long period of time
give up their aspirations to compete and excel in an activity.
Without the goal of improving performance, the motivation to
engage in practice vanishes. Kaminski, Mayer, and Ruoff
(1984) found that many elite adolescents who decided to stop
competing remained active in the domain but virtually stopped
engaging in practice.
Some individuals have had to terminate their professional
careers for reasons unrelated to their ability to perform. In a
longitudinal study of visual artists, Getzels and Csikszentmiha-
lyi (1976) found that most artists were drawn to painting be-
cause it allowed social isolation. However, aspiring painters
have to promote social relations with art dealers, art critics, and
buyers to gain notoriety, increase the demand for their art, and
generate sufficient sales for full-time artistic activity. Failure to
do so forced many of the best artists to take another job unre-
lated to painting. Once these artists could no longer commit
sufficient time and energy to maintain and improve their per-
formance they stopped painting completely because they could
not accept performing at a lower level. This finding shows that
the activity of painting as such is not inherently motivating but
rather the act of producing art that satisfies the artists' subjec-
tive criteria for quality.
Implications for Empirical Studies
Applied to an individual, our theoretical framework made
three types of predictions: (a) predictions about the develop-
mental history; (b) predictions about current levels and habits
of practice; and (c) predictions about experts' evaluations re-
garding the nature and role of deliberate practice activities that
are relevant throughout development.
Our framework made two important predictions about an
individual's development history. First, the past amount of de-
liberate practice is directly related to the individual's current
performance. More specifically, expert performance is not
reached with less than 10 years of deliberate practice. Second,
deliberate practice starts at low levels and increases slowly over
time. These predictions can be best tested in domains of exper-
tise that are relatively independent of the traditional school
system and where deliberate practice can be easily identified
and measured. It is important that the domain has qualified
teachers who guide individuals in learning basic skills correctly
and direct them toward optimal practice activities. Music is one
such domain, and in the current studies we chose to study
individuals who perform at very high levels on a particular
instrument. To study individuals who had completed the 10-
year period of preparation and had made a commitment to
music as a profession, we contacted the Music Academy in West
Berlin. This academy has an international reputation for its
training program for violinists. Violinists in this program were
asked to provide retrospective reports on their levels of deliber-
ate practice over the years before they entered the academy so
that we might test our predictions.
Next, our framework made several predictions regarding the
current level and related habits in elite performers. First, the
highest improvement of performance, and indirectly the high-
est attained performance, is associated with the largest weekly
amounts of deliberate practice. We predicted that elite per-
formers practice at a constant level from day to day to maxi-
mize improvement over extended periods of time. Further-
more, the daily periods of deliberate practice should be of lim-
ited duration with rest periods in between. In domains with
weekly competitions, stability and equilibrium should occur
over longer time periods such as a week. The extremely effortful
nature of competition, normally on the weekend, would lead to
a reduced load and duration after and just before the competi-
tions.
To obtain information about music performers' current prac-
tice patterns, we asked them to keep diaries. By collecting de-
tailed diaries by these individuals, we could assess the duration
and regularity of different types of activities, in particular those
activities judged to constitute deliberate practice. Drawing on
earlier research on time budgeting (Juster & Stafford, 1985;
Szalai, 1972), we had individuals record at the end of the day all
extended activities with their start and end times. By recalling
already completed activities, individuals who maintain this
diary report should minimize any biasing influence on the fre-
quency and duration of any activity during the day. Further-
more, the instruction to recall the complete sequence of ex-
tended activities during the entire day avoids the bias of focus-
ing on a single activity. This type of diary report is consistent
with Ericsson and Simon's (1984) criteria for valid and unbiased
verbal reports of cognitive processes. This technique is prefera-
ble to an alternative in which subjects are instructed to keep a
selective diary for occurrences of specific problem behaviors,
such as drinking and smoking; the keeping of such a diary
appears to reduce the frequency of these behaviors and thus
yields biased estimates (Hodgson & Miller, 1982).
THE ROLE OF DELIBERATE PRACTICE
373
Large-scale studies have evaluated the accuracy and conver-
gent validity of diary reports by comparing subjects' diary esti-
mates with estimates derived from random time sampling (Ro-
binson, 1985). The diaries were found to underreport activities
of very short duration, such as brief social interactions and
phone calls, a result that is to be expected with diaries focusing
on extended activities. More important for our purposes, the
diary estimates for extended activities were found to be quite
consistent with the results derived from more labor-intensive
methods.
Most of the research using diaries with reported temporal
sequences of activities has been conducted primarily in socio-
logical and economic studies to estimate and project the use of
time in representative national populations (Juster & Stafford,
1985; Szalai, 1972). The goal of this research has been to derive
general categories of activities that allow investigators to reli-
ably classify any one of the reported activities into one of a
limited number of categories. At the highest level, the activities
can be grouped into categories, such as sleep, work, and leisure.
In studying the daily lives of expert performers, we can draw on
this previously developed classification for general activities,
but we must supplement it with an analysis of the activities
relevant to the particular domain of expertise under investiga-
tion.
As to the third point, our framework made predictions about
the qualities of various domain-related activities, such as delib-
erate practice. We predicted that deliberate practice would be
rated very high on relevance for performance, high on effort,
and comparatively low on inherent enjoyment. We could evalu-
ate ratings by expert individuals to determine the extent to
which deliberate practice is perceived to have these attributes.
The central prediction from our framework was that the
adult elite performance, even among individuals with more
than 10 years of practice, is related to the amount of deliberate
practice. This prediction contradicts Galton's (1869/1979) mo-
dal view outlined earlier that eminent performance reflects pri-
marily innate talent after sufficient practice and that, by impli-
cation, practice and elite performance are not related. However,
often talent is contrasted with practice, where the best individ-
uals are assumed to practice less than individuals with inferior
performance. Finally, plausible alternative hypotheses also sug-
gest that the most talented individuals would practice more.
These hypotheses imply a high correlation between innate tal-
ent and practice. Because our studies were not designed to ad-
dress the last possibility, we do not consider it except in the
General Discussion section.
Study 1 compared the current and past levels of practice in
three groups, elite violinists judged to have promise for careers
as international soloists and two groups of less accomplished
expert violinists. Study 2 replicated the results of the first by
comparing expert and amateur pianists. In addition, it related
estimates of the amount of prior practice to current perfor-
mance on a wide range of musical and nonmusical tasks for all
pianists.
Study 1
We assessed current and past levels of deliberate practice in
three groups of elite, adult violinists whose current perfor-
mance differed. First we identified the activities constituting
deliberate practice. We then determined the duration and orga-
nization of deliberate practice and contrasted them for the
three groups.
Method
Subjects
The music professors at the Music Academy of West Berlin (Hochs-
chule der Kuenste) nominated violin students who had the potential for
careers as international soloists. Out of 14 students nominated, 3 were
not fluent in German and 1 was pregnant. The remaining 10 students
agreed to participate in the study and are called "the best violinists."
The music professors also nominated a large number of good violinists
in the same department. From these subjects, we selected 10 violinists,
"the good violinists," by matching their sex and age to those of the best
violinists. Similarly, we recruited 10 students specializing in the violin
from a different department (music education) in the academy, which
has lower admissions standards. Again, we matched these students' sex
and age to those of the violinists in the best group. We call the students
from the department of music education the "music teachers" because
teaching is the most likely future profession for this group. To obtain
additional data on the developmental history of outstanding violin-
ists, we interviewed 10 middle-aged violinists who belong to two sym-
phony orchestras in West Berlin with international reputations, the
Berlin Philharmonic Orchestra and the Radio Symphony Orchestra
(RSO). According to the professors at the music academy, the most
likely professional career for the best young violinists is to perform as
a member of one of the best symphony orchestras in Germany.
Procedure
The data-collection procedures for the best and good violinists and
the music teachers were identical. The first part of the procedure for
the middle-aged violinists, from which data are reported, was the same
as for the three groups of young violinists.
Each young violinist was interviewed during three sessions. During
the first session biographical information was obtained including the
start of practice, sequence of music teachers, and participation in com-
petitions. The subjects were then asked to estimate how many hours
per week they had practiced alone with the violin for each year since
they had started to practice.
Each subject was then given instructions about a taxonomy of activi-
ties. Ten categories of everyday activities were presented, each with a
general label and description and a listing of frequent examples. From
extensive pilot work, 12 categories of musical activities were identified
and similarly presented. Musical and everyday activities are listed in
Table 1. For those subjects playing other instruments besides the vio-
lin, the eight musical categories for playing an instrument were split
into activities involving the violin and activities involving all other
instruments. After presentation of the taxonomy, subjects were asked
to estimate how much time they spent on each type of activity during
the most recent typical week. Subjects were also asked to rate each of
the activities on three dimensions using a scale from 0-10. First they
were asked to rate the relevance of the activity to improving perfor-
mance on the violin. Next they were asked to rate the effort required to
perform the activity. Finally, they were asked to rate how enjoyable the
activity was without considering their evaluation of the result of the
activity. (For example, it is possible to enjoy the result of having cleaned
one's house without enjoying the activity of cleaning.)
During the second session, subjects answered questions about prac-
tice and concentration. They also recalled all activities they had en-
gaged in during the previous day. For this recall they used a specially
374
K. ERICSSON, R. KRAMPE, AND C. TESCH-ROMER
Table 1
Mean Relevance, Effort, and Pleasure Ratings for 12
Music-Related and 10 Everyday Activities Collapsed
Over Three Groups of Yo ung Expert Violinist s
Activities Relevance Effort Pleasure
Music related
Practice (alone) 9.82 H 8.00 H 7.23
Practice
(
with others
)
8.73 H 6.97 H 7.57
Playing for fun
(
alone
)
5.67 3.27 L 8.33 H
Playing for fun
(
with others
)
6.67 3.93 8.60 H
Takin
lessons 9.63 H 8.60 H 7.67
Giving lessons 7.03 7.51 H
a
6.79
a
Solo
p
erformance 9.03 H 9.80 H 7.28
a
Group performance 7.67 H 8.14 H
a
8.07 H
Listenin
g
to music 8.33 H 4.38
a
8.38 H
Music theory 7.63 H 6.37 H 6.07
Professional conversation 6.50 4.33 6.40
Organization and
p
re
p
aration 2.90 L 4.70 1.53 L
Everyday
Household chores 1.80 L 2.23 L 3.63 L
Child care 2.64 L
b
6.14
b
6.43
b
Shopping 0.77 L 2.80 L 3.97 L
Work (not music related) 1.79
L
b
5.5
6
C
3.74 L
c
Bod
y
care and health 4.90 1.43 L 5.23
Sleep 8.17 H 0.47 L 7.70
Education (not music) 4.52
a
5.45
a
7.17
a
Committee work 1.93
L
a
5.55
a
5.07
a
Leisure 6.30 3.00 L 8.93 H
Sports 6.07 2.67 L 7.07
Grand mean 5.89 5.03 6.52
Note. N = 30, unless shown with a superscript. (Some subjects could
not make their ratings because of a lack of familiarity with the activity
in question.) The grand means over all activities and information about
the significant deviation from the grand mean is given based on post
hoc analyses using Bonferroni's method. The statistical test is conser-
vative as the grand mean includes the ratings for the particular activity
in the respective comparison. H = significantly higher than grand
mean; L = significantly lower than grand mean.
a
N=29.
b
N=28.
C
N=27.
designed diary sheet that divided the 24-hr day into ninety-six 15-min-
ute intervals. Use of the sheet ensured that the start and end of recalled
activities covered the entire 24-hr day. After completing the recall,
subjects were asked to encode the activities using the 30 categories in
the taxonomy. Following the second session, subjects kept a diary us-
ing the provided sheets for a full 7-day week. Subjects were given enve-
lopes addressed to the investigators and sent in their diaries after each
day. Before returning for the third and final interview session, the
subjects, working from copies of their diaries, encoded each activity
according to the taxonomy. Subjects were encouraged to identify the
primary category for each activity but they were allowed to use more
than one category to encode mixtures of activities, such as a profes-
sional discussion during lunch. At the beginning of the third interview
session, subjects were allowed to ask any questions they had about
their encoding. During the remaining part of the session the inter-
viewer asked questions about the subjects' developmental life goals
and engaged in general debriefing.
Results
Our analyses focus on young violinists' allocation of time for
relevant preparatory activities as revealed by their diaries and
retrospective estimates of practice during their development. In
preparation for this analysis, we briefly describe the biographic
data and other data relevant to systematic differences in the
violin performance of the three groups of young violinists and
analyze each group's ratings of everyday and musical activities
regarding relevance, effort, and enjoyment.
The three groups of subjects were selected such that the per-
formance of the best violinists should be better than that of the
good violinists, whose performance in turn should be better
than that of the music teachers. In the following statistical anal-
yses, the hypothesized differences between the three groups of
violinists are represented by two orthogonal contrasts. The first
contrast refers to the average difference between the best and
the good violinists. The second orthogonal contrast compares
the average of the best and good violinists (referred to as the
soloist students) to that of the music teachers. In statistical anal-
yses that include the developmental history of middle-aged pro-
fessional violinists, the data from this group are contrasted
with those of the group of the best young violinists.
Biographic Information
All three groups of young violinists consisted of 7 women
and 3 men. The middle-aged professional violinists were all
men. The ages of the young subjects were successfully matched,
and no reliable differences in age were found. The mean age of
young violinists was 23.1 years old. The mean age of the profes-
sional violinists was 50.5 years old.
The biographic histories of the four groups of subjects with
respect to violin playing are remarkably similar and show no
systematic differences between groups. The age when they be-
gan practice was 7.9 years old and essentially coincided with
the age of starting systematic lessons, which was 8.0-years-old.
The age at which they first decided to become musicians was
14.9 years old. The average number of music teachers who had
taught them was 4.1, and the average number of musical instru-
ments that they had studied beyond the violin was 1.8.
The best indicator of violin performance, besides the evalua-
tion of the music professors, is success at open competitions. A
statistical analysis of the number of successful entries in violin
competitions confirmed systematic differences in performance
among the three groups of young violinists. The frequencies for
the best and good violinists were reliably different, 2.9 vs. 0.6;
F(\, 27) = 19.35, p < .01. The average frequency of the best and
good violinists differed from that for the music teachers, 1.8 vs.
0.2; F(i, 27) = 11.78, p < .01. The same pattern of results
emerged when the proportion of successful entries from all
participations was analyzed. The young violinists were also
asked to estimate in minutes of playing time how much music
they could perform from memory without preparation. The
best violinists reported an average of 128.9 min, which is longer
than the 79.1 min reported by the good violinists, F(l, 27) =
4.07, p < .05. The average playing times for the best and good
violinists were longer than that of the music teachers, 104.0 vs.
42.27; F(l, 27) = 8.23, p < .015.
In sum, all four groups had a similar musical background,
and by the age of 23 (the mean age of the young violinists), all
40 subjects had spent at least 10 years practicing the violin.
THE ROLE OF DELIBERATE PRACTICE
375
Ratings of Everyday and Musical Activities
In analyzing the ratings of relevance to improving violin per-
formance, effort, and enjoyment of the everyday and musical
activities, our primary goal was to identify a smaller set of activi-
ties rated critical to improvement of violin performance by all
young violinists. Analyses of each set of ratings for the 22 activi-
ties for the three groups of young violinists revealed no profile
differences, that is, interactions between group and ratings of
activities, which could account for differences in their actual
allocation of time to different types of activities. We therefore
collapsed the further analyses of differences in ratings between
various activities across the three groups of young violinists.
For each type of rating (relevance, effort, and enjoyment) we
compared the mean rating across all activities with the mean
rating of that activity. The significance of differences was deter-
mined from adjusted alpha levels (a* = a/22 = 0.0023). The
mean ratings and information about significant differences are
given in Table 1.
As shown in Table 1, subjects reliably rated 7 of the 12 musical
activities as more relevant than the overall mean. Consistent
with our theoretical assumption, 27 of the 30 violinists gave
"practice alone" the highest relevance rating. In contrast, play-
ing alone for fun, which an observer would have difficulty dis-
criminating from practice alone, received a much lower rele-
vance rating. Out of the 10 everyday activities, only sleep was
rated as reliably more relevant to improving one's violin perfor-
mance than the grand mean, and 5 activities were rated reliably
less relevant. Many of the activities with the highest relevance
ratings are constrained by external factors and resources. For
example, the duration of taking lessons and public perfor-
mance alone and in a group cannot easily be increased at the
will of the subject. Similarly, practice in groups is to a lesser
degree constrained. After these additional criteria are applied,
there remain four relevant activities of which the violinists can
easily control the duration: practice alone, music theory, listen-
ing to music, and sleep.
The ratings of effort associated with different activities show
that six of the eight activities judged to be highly relevant to
performance improvement are also judged to require reliably
more effort than the average activity. The two exceptions are
listening to music and, not surprisingly, sleep, which is judged
reliably less effortful than the average activity. The ratings of
inherent pleasure show that only two of the eight highly relevant
activities—listening to music and group performance—are
also judged to be reliably more pleasurable than the average
activity.
In sum, all three groups seem to have the same conception of
the relevance of different activities for improvement of violin
performance, and all three similarly evaluate the inherent en-
joyment and effort associated with different activities. We now
turn to the analysis of the time each group allocated to different
activities and, in particular, to activities judged to be highly
relevant for improving violin performance.
Diaries
From the detailed diaries with encoded activities, the total
time a violinist spent during the week on any one of the activity
categories can be calculated by simple addition. Most of the
encoded activities were described by a single category, such as
exclusive categories like sleep and practice; but some activities,
such as a professional discussion over dinner, were given multi-
ple encodings. When activities were given multiple encodings,
the time of the activity was split equally among the associated
categories.
When the duration of all music-related activities was
summed across the diary week, the average number of hours
per week was 50.6, and no reliable differences between the
groups were found. Of the eight activities judged to be highly
relevant to improvement of violin performance, only two had
an average duration across all three groups exceeding 5 hr per
week. These two activities were practice alone (19.3 hr per
week) and sleep (58.2 hr per week).
Practice alone. In agreement with our theoretical frame-
work, violinists rated practice alone as the most important activ-
ity related to improvement of violin performance. Practice
alone is a particularly interesting activity because the violinists
themselves control its duration and distribution during the
week. In contrast, most other activities judged to improve vio-
lin performance, such as public solo performance and taking
lessons, are highly constrained by external factors. We analyzed
the total duration of practice alone for the three groups and
then examined the distribution of practice alone during the
week.
During the diary week, the average duration of the violinists'
practice alone with the violin did not differ for the two best
groups and averaged 24.3 hr of practice. This average was reli-
ably greater than that for the music teachers who practiced 9.3
hr per week, F(l, 27) = 44.05, p < .001. As the first step in
analyzing the distribution of practice, we analyzed the daily
amount of practice as a function of the day of the week for the
three groups. No main effect or interaction of the day of the
week was observed, and only the contrast found earlier between
the two best groups and the music teachers was reliable. The
two best groups practiced alone for 3.5 hr per day and the music
teachers for 1.3 hr per day for each day of the week including the
weekend. As the second step, we assessed the frequency of
practice as a function of time of day. The frequency distribu-
tions across all weekdays shown in Figure 2 suggest a prefer-
ence by the two best groups for practicing alone before lunch,
whereas no corresponding pattern is observed for the music
teachers.
For statistical analyses, the percentage of time each violinist
spent practicing alone was calculated for the five 2-hr intervals
from 10:00 am to 8:00 pm for each day of the week. An analysis
of variance (ANOVA) revealed no main effects or interactions
involving the day of the week and no systematic differences
between the two best groups. The two best groups spent a
greater proportion of time on practice alone than the music
teachers did, F(l, 27) = 59.11, p < .001, and this difference
interacted with the time of day, F(4,108) = 2.94, p < .05. A post
hoc analysis showed that the time music teachers practiced
alone was distributed uniformly across the day, whereas the two
best groups had elevated levels of practice between 10:00 am
and 2:00 pm. This interaction and the main effect of time of
day, F(4,108) = 6.09, p < .001, are shown in Figure 3.
Consistent with the rated effortfulness of practice alone, the