Background The increasing prevalence of overweight and obesity in childhood has implications for their future health. There are many potential contributors to overweight and obesity in childhood. The aim was to investigate the association between postulated risk factors and body mass index (BMI) in children and adolescents. Methods Secondary analysis of data from a multi-centre, multi-country, cross-sectional study (ISAAC Phase Three). Parents/guardians of children aged 6?7 years completed a questionnaire about their child?s current height and weight, and the postulated risk factors. Adolescents aged 13?14 years reported their own height and weight and answered questions about the postulated risk factors. A general linear mixed model was used to determine the association between BMI and the postulated risk factors. Imputation was used if there were missing responses for 3 or fewer explanatory variables. Results 65,721 children (27 centres, 15 countries) and 189,282 adolescents (70 centres, 35 countries) were included in the final analyses. Many statistically significant associations were identified, although for most variables the effect sizes were small. In children birth weight (for each kg increase in birth weight the BMI increased by +0.43 kg/m2, p<0.001), television viewing (5+ hours/day +0.33 kg/m2 vs. <1 hour/day, p<0.001), fast food (?3 times/week +0.16 kg/m2 vs. never, p<0.001) vigorous physical activity (3+ hours/week 0.071 kg/m2 vs. never, p = 0.023) and maternal smoking in the first year of life (+0.13 kg/m2, p<0.001) were associated with a higher BMI in the adjusted model. Nut consumption (?3 times/week -0.11 kg/m2 vs. never, p = 0.002) was associated with a lower BMI. Early life exposures (antibiotics, paracetamol and breast feeding) were also associated with BMI. For adolescents statistically significant associations with BMI and were seen with maternal smoking (+0.25 kg/m2, p<0.001), television viewing (5+ hours/day +0.23 kg/m2 vs. <1 hour/day, p<0.001), fast food (?3 times/week -0.19 kg/m2 vs. never, p<0.001), vigorous physical activity (3+ hours/week 0.047 kg/m2 vs. never, p<0.001) and nuts (?3 times/week -0.22 kg/m2 vs. never, p<0.001). Conclusions Although several early life exposures were associated with small differences in BMI, most effect sizes were small. Larger effect sizes were seen with current maternal smoking, television viewing (both with higher BMI) and frequent nut consumption (lower BMI) in both children and adolescents, suggesting that current behaviours are more important than early exposures. Although many variables may influence BMI in childhood, the putative factors studied are not of sufficient magnitude to support major public health interventions.