Forty controlled studies that provided 66 comparisons (n=31,059) were included.
There was a small statistically significant benefit with school-based interventions compared to control (r=0.05, 95% CI 0.04 to 0.06); there was statistically significant heterogeneity in this analysis (p<0.001). There were statistically significant between-group differences for several of the moderator variables. More successful interventions were those that were universal (r=0.07, 95% CI 0.05 to 0.08; 37 comparisons), undertaken with four to nine year olds (r=0.06, 95% CI 0.05 to 0.07; 41 comparisons), implemented collaboratively (r=0.12, 95% CI 0.10 to 0.14; 19 comparisons), primarily included children of Asian ethnicity (r=0.30, 95% CI 0.27 to 0.33; five comparisons), encouraged nutritional change (r=0.13, 95% CI: 0.11 to 0.14; 28 comparisons) and sought to reduce sedentary behaviours (r=0.15, 95% CI 0.13 to 0.17; 17 comparisons).
There was a negative effect for short-duration interventions (r=-0.04, 95% CI -0.08 to 0.00; 11 comparisons) and those that implemented system-wide changes in nutrition (r=-0.03, 95% CI -0.04 to -0.01; 15 comparisons).