The databases searched included MEDLINE (1966 to 2007), PSYCHINFO (1974 to 2007), SCHOLAR.GOOGLE.COM, and ERIC, as well as the extensive personal databases of the authors. The reference lists of the articles thus identified were also consulted to identify additional potentially-relevant research. Publications in languages other than English were considered where appropriate. For the purpose of this review, we use the term academic achievement to encompass academic success, school performance and all combinations of these terms.

The outcomes of school PA/PE and academic achievement, success or performance were actual or self-reported grade point average (GPA) and determinants of GPA that could potentially be changed by the interventions (concentration, learning, classroom behaviour, engagement in learning, self-esteem, etc.). The terms physical education, physical activity and sports are, for the purposes of this review, restricted to programmes offered within the school context (i.e. instructional physical education and extracurricular physical activity, including in-school physical activity programmes, intraschool and intramural sport).

It is not surprising that no randomized controlled trials were identified, as they are not practicable in this type of research setting. Quasi-experimental protocols are usually indicated when causality cannot be tested by a random controlled trial in milieux such as the school setting. Seven quasi-experimental studies were identified (Table 1). Cross-sectional studies were also considered, as well as experimental or laboratory experiments on the determinants of academic performance (i.e. learning concentration, classroom behaviour, etc.).

The first documented quasi-experimental study relevant for to this paper was the Vanves (France) investigation; this involved a small group of schoolchildren tested during the 1950's. 7 Schoolchildren who spent mornings in the classroom and afternoons doing PE were said to perform better academically than children from a control class, but no further details were given. 7 Unfortunately, the specifics of these observations were not described in peer reviewed journals.

A second quasi-experimental study conducted in the Trois-Rivières region (Québec) between 1970 and 1977 involved 546 primary school students; this noted that students involved in an experimental 5 hours of physical education per week had a higher academic performance than their control counterparts who were enrolled in the normal school program for 40 min per week 8. The supplemental 260 minutes allocated to PE was necessarily taken from time for other academic teaching (i.e. an average 14% curtailment of academic instruction). Despite this curricular change, during the last 5 years of primary school, the overall academic performance of the experimental students improved relative to the controls. During standardized Provincial examinations, children receiving the 5 hours/week of PE had higher scores in mathematics, but lower scores in English (their second language), despite the fact that 33 minutes were removed from mathematics instruction and none from English. 3

A 2-year quasi-experimental study followed 759 Californian children in the 5th and 6th grades. 9 Subgroups of children were taught PE by either a professional physical educator (n = 178), a trained homeroom teacher (n = 312), or in the normal programme (n = 165). The professional physical educators, the trained teachers, and normal programmes offered, respectively, 80, 65, and 38 minutes per week of PE. As expected, those taught by the professional physical educators achieved greater fitness (cardiovascular and muscle endurance). 10 Also, the groups taught by the professional physical educators and trained teachers had smaller declines in academic performance despite allocating more time to PE. Four of 8 statistical comparisons disclosed an advantage for students in the experimental groups; one comparison was advantageous to control students, while the remaining 3 were equal. The group who spent the most time on PE (i.e. those with a professional physical educator) showed no negative effects on academic achievement and indeed the decline of academic results during the 2 years of the intervention was smaller than that observed in the control subjects. 9

In South Australia, the 500-student SHAPE trial added 1.25 hours per day of endurance fitness training to the curriculum of 10-year-old primary school students. 11 Over the first 14 weeks of the study, the experimenatl group showed gains in physical work capacity and decreases in body fat relative to controls. Arithmetic and reading scores were not adversely affected by the substantial reallocation of curricular time in favour of PE. These physical benefits appeared to be maintained over the succeeding 2 years in a follow-up of 216 participants. These follow-up evaluations showed (non-significant) trends for better arithmetic and reading grades in experimental students, as well as beneficial changes in teachers' ratings of classroom behaviour. 12

The 16-month Action School BC! project involved a population of 287 British Columbian primary school children (4th and 5th years: 9–11-years olds). PA was delivered by classroom teachers, amounting to 47 minutes more per week in interventional than in control schools (139 ± 62 vs. 92 ± 45 minutes, P < 0.001). 13 Despite a corresponding decrease in academic time, the academic performance of the experimental group, as measured by the Canadian Achievement Test, remained unchanged; indeed, data analysis revealed a trend towards an enhanced academic performance in the intervention schools (the average score rising from 1,595 to 1,672 units).

Another interventional study of 6^th grade (11 year-old) students covered a single school term. Fifty-five minutes/day of PE were included in the curriculum, vs. the same allocation of time for arts or computer sciences; the two groups performed equally well in mathematics, sciences and English. 14 Finally, an intervention in Israel involved 92 preschool and 266 first grade children. 15 The experimental manipulation here was a school-based movement education programme, and children in the experimental group showed greater reading skills and arithmetic scores than controls. 15

Taken together, these quasi-experimental data suggest that the enriched PE programmes demanded a substantial reduction in the time allocated for academic tuition. Since the children achieved at least equally despite the reduced teaching time, the evidence seems strong that the efficiency of learning was enhanced. 3 Despite the variety of programme durations and locations, a common and valuable by-product was a significant increase in various measures of physical fitness (PF).

Cross-sectional studies commonly have difficulty in controlling for potential biases, particularly socio-economic status (SES). SES remains the strongest predictor of academic achievement 16 and is also one of the strongest predictors of PA participation in children (e.g. in Canada 17; Italy 18 and Estonia 19). Cross-sectional studies generally indicate a positive association with academic achievement. Some of these studies did control for confounders such as SES, and still most of them found a positive association between physical activity and academic achievement (Table 2).

Many factors like classroom behaviour, self-esteem, self-image, school satisfaction and school connectedness have been postulated as determinants of academic achievement.

Many authors have documented the acute effects of PA on cognitive function. Three recent reviews and/or meta-analyses examined these studies. 444546 In a meta-analysis of 44 studies, Sibley and Etnier 45 concluded that PA was positively associated with better cognitive functioning in children. Some groups, particularly middle school students (grades 6–8, aged 11–13 years) and younger, seemed to benefit more from PA. Sibley and Etnier 45 noted that unpublished studies had a higher effect size than published reports, suggesting that no bias had occurred from a failure to publish non-significant results.

Brisswalter et al. 44 reviewed published studies into the effects of exercise on various tasks. They concluded that the optimal intensity for decisional tasks covered a wide range (~40–80% VO₂max). An exercise duration of more than 20 minutes was most efficient in increasing the performance of perceptual and decisional tasks. 4446 Tomporowki 47 suggested an upper limit of 60 minutes might arise from the adverse effects of dehydration on cognitive functions.

The literature generally suggests a positive effect of acute physical exercise on cognition. Other activities, like involvement in music also have the potential to increase reading skills, although in this case there is no positive influence on PF. 48

What is the effect of a high level of PF on academic performance? Is good cognitive functioning associated with above average PF? If so, is this a consequence of PF per se, or of better overall physical health? When analyzed globally, the literature does not indicate any clear linkage between PF and either academic achievement or intellectual performance. As early as 1969, Railo found no relationship between PF and either of these outcomes. 49 More recently, Etnier et al. 50 concluded from a meta-regression analysis that the empirical literature did not support a link between cardiovascular PF and academic achievement. However, this meta-analysis revealed a weakness in the literature: there was little data on the relationship between PF and academic achievement in school-aged children. Indeed, only 1 of the 37 studies identified included this age group.

When the definition of PF includes aspects other than cardiovascular fitness, there seems evidence of positive correlations between various measures of psychomotor performance, cognitive abilities and academic achievement. 5152 Psychomotor performance shares many common neurological mechanisms with cognitive functions.

A 2001 cross-sectional study on California children disclosed a positive relationship between reading and mathematics results (as measured by Stanford Achievement Test-9) and results on a field test of physical fitness (the Fitnessgram). Despite a huge sample of students from grades 5, 7 and 9 (n = 954,000), potential selection biases were not considered, making it difficult to conclude that PA was linked to increased academic performance. 53 When found, any effects of PF were small. Another weak association between PF and academic achievement was observed in South Korean children (grades 5, 8, and 11); in this study, the association was much smaller than that between academic achievement and regular meal eating. 54 Dwyer et al. 24 measured muscle fitness in 9,000 Australian students. They found significant but weak associations, ranging from r = -0.10 to -0.19 for running distances of 50 m and 1.6 km, and from r = 0.10 to 0.22 for sit-ups and standing long jump, respectively.

The connection between school sports and intellectual achievement has been a long-standing issue since Davis and Cooper 55 first reported a positive association between school sports participation and academic achievement. It remains the subject of recent investigations. The competitive dimension of most sports introduces particular problems, even in the school context, as the educational dimension tends to be relegated to a secondary level. The literature comprises mainly cross-sectional data and the results are more equivocal than for PA; unfortunately, most of the earlier studies did not control for biases common to athletic and academic achievements. 5657

Data from the longitudinal Maryland Adolescent Development in Context Study included 67% African-Americans and 33% European-Americans; it found that participation in extracurricular PA was a significant predictor of better academic results and of higher academic expectations. 58 Furthermore, sports participation by 8th grade African-American males resulted in aspirations to continue their studies toward college, with less likelihood of acting inappropriately in school. 59 In their female counterparts, sports participation also resulted in higher aspirations and in a reduction of absenteeism.

Cooper et al. 60 found that even after eliminating confounding factors, extracurricular activities, including sports and PA were predictors of better academic achievement in 2,200 American high school students. Their conclusion is in line with the point that Marsh made in 1992, that such activities may have an effect on academic achievement by increasing motivation and investment in school. 61 Another study of 11,957 American adolescents found that even after standardization for SES, sports participation with parental presence was associated with an increased probability of good grades in English and mathematics, the Adjusted Relative Risk being 1.23 for both subjects. 20 Dexter 62 examined the relationship between sports knowledge, sport performance and academic ability, the last being measured by scores on the British General Certificate of Secondary Education (GCSE). They observed a small but significant positive correlation between sports performance and GCSE score for both mathematics and English.

Melnick et al. 63 detected no relationship between academic achievement and sports participation in 3,686 African-American and Hispanic students from the "High-school and Beyond Study". However, sports participation was associated with a lower drop-out rate. Therefore, they suggested that if sports participation contributes to academic achievement, it may do so indirectly, by encouraging retention in school. Fisher et al. 64 also observed no association between sports involvement and self-reported grades in an ethnic mix of 838 grade 9 to 12 students (predominantly 63% African-American and 27% Hispanic).

Harvard students involved in varsity teams had a slightly lower GPA than their peers, but reported a higher degree of satisfaction with their university experience. 65 This also seemed the case in other institutions examined by Light. Athletes have more friends and a stronger sense of belonging to their institution. They are, according to Light, "the happiest on campus". Generally, this same trend is seen among high-school athletes. Students engaged in extracurricular PAs do not achive different academic scores than their peers, but they feel a greater engagement with their institution. 6667 This may reflect in part the greater attention directed towards these specific students. Indeed, participants in extracurricular activities (including sports) have more interactions with significant adults than non-participants. 66

Sport is a very complex phenomenon. There are many cultures within school sports, and any effect on academic achievement is influenced by gender, race, type of sport, type and level of athletic involvement. White and McTeer 68 suggested that the status of a given sport may influence its effect on academic achievement. Their results showed that high-status sports had a positive influence on English grades but they saw no evidenceof an effect of such sports on mathematics grades. They suggested that academic performance was more likely to be affected by cultural factors in subjective subjects like English than in mathematics. Any influence of school sports participation may also differ between girls and boys 33, and between various ethnic and cultural groups. 69

In conclusion, the available literature suggests that sport is more likely to benefit academic achievement if offered in school rather than in other sport contexts, given the proximity of educational resources and environment. This may be particularly important for team sports, which often seem associated with risky behaviours, particularly binge drinking of alcohol. 70 When sports-involved students identify themselves as athletes rather than «jocks», such risky behaviours seem less prevalent. 67 Greater academic coaching of school athletes could be a factor favouring their academic achievement. 67 School sports should be monitored closely, with the intent of avoiding a drift away from educational objectives. It appears that satisfaction with sports vs. satisfaction with school work is predicted by a differing psychological domain (perceived ability vs. task orientation). 71 It may be helpful to create an environment where both types of endeavour find common ground, i.e. school may be the best setting in which sports can be directed towards task orientation and skills acquisition, without decreasing the pleasure and satisfaction of being good at sports and PA. As noted in various long-term follow-ups, elite and varsity level athletes later tend to experience greater educational and labour market success than non athletes. 34677273 Current evidence suggests that this effect may be mediated by racial group. 74

Academic integration of children with various behavioural and developmental problems is a growing trend in industrialized countries. The question arises in terms of their academic achievement. Reviews of exercise programmes for children with learning disabilities 7576 have suggested that in order to increase the likelihood of positive outcomes, such programmes should have a low student-instructor ratio. Benefits (with the exception of increased PF) may reflect increased attention toward the participants.

In hyperactive impulsive children, PA is associated with global satisfaction in boys and an absence of depressive emotions in both sexes. 77 An outdoor education programme also decreased behavioural problems in children with attention deficit hyperactivity disorder. 78

In children with reading disabilities, a school-based programme of balance and coordination training, throwing, catching, and stretching produced significant improvements in both reading and semantics. 79 Positive changes were maintained for at least 18 months following the programme, reducing the likelihood of a Hawthorne effect. 80

Four pupils with emotional and behavioural disorders were directly studied before and after a 10-week PE intervention. Back in class, there was an increase (13.8%, or a little more than 23 minutes) in the amount of time spent focused on the tasks they were supposed to be performing. 81 A 10-week PA intervention in children with learning disabilities improved classroom behaviour and the perception of academic competence was increased. 76 However, a similar outcome was seen in the control group, indicating that there had been no specific effect from the programme.

The effects of school PA upon children with learning problems thus remains an open field for research.

The positive association observed between PA and intellectual performance among children in quasi-experimental studies should be supported by mechanistic, experimental evidence. No one can deny the important role of neurosciences in the comprehension of academic achievement. 82 Most research on the relationships between PA and cognition has centered on the hippocampus, a brain region that mediates memory and learning in mammals, and on changes in the cerebral circulation. The hippocampus has an important role in the consolidation of memory. One major mechanism essential to its functions is long-term potentiation, or LTP. LTP leads to an enhancement of nervous influx following a first series of stimuli.