Background Childhood obesity has been shown to affect human capital and social outcomes later in life. Yet, evidence of the causal nature of this link is scarce and pathways are not well understood. We aimed to investigate the effect of childhood obesity on cognitive performance in adolescence and educational attainment in early adulthood.

Methods We used data of three longitudinal UK cohorts: the1958 National Child Development Study (NCDS; n=5346), the 1970 British Cohort Study (BCS70; n=6790) and the Avon Longitudinal Study of Parents and Children (ALSPAC; n=5373) which includes children born in1991–92. We used ordinary least squares and logistic regression, value-added sex-stratified models, and mendelian randomisation models to explore the effect of childhood body-mass index (BMI; Z score and BMI category at age 11 years or at 16 years) on cognitive performance (Maths and English scores at age 16 years) and educational attainment (tertiary qualification at age 23 years).

Results In ALSPAC, female individuals who had been overweight at age 11 years scored 1·00 point (95% CI –1·58 to –0·36; p=0·028) less on their maths GCSE exam than their healthy-weight peers, and girls who had been obese at age 11 years scored 1·66 points (–3·15 to –0·18; p=0·0021) less. Female individuals who had been obese at age 11 years were less likely to graduate from university than their healthy-weight peers (odds ratio [OR] 0·75, 95% CI 0·59 to 0·97; p=0·030); the effect on graduation for those who had been overweight was less conclusive (OR 0·85, 0·71 to 1·01; p=0·060). Male individuals who had been overweight scored 1·21 points (95% CI –2·84 to –0·81; p=0·0011) less and those who had been obese 2·24 points (95% CI –3·46 to –1·02; p<0·0001) less on their GCSE maths exam than their non-obese peers, but there was no association between male childhood weight and university graduation (overweight: OR 1·07, 0·95 to 1·21; p=0·26; obesity: OR 0·89, 0·66 to 1·21; p=0·47). In BCS70 and NCDS, there was a positive but insignificant relationship between overweight/obesity and cognitive performance but no significant findings for educational attainment.

Conclusion Our findings are robust to various causal methods and might help inform interventions to address this issue. Cross-cohort comparisons suggest that there might be a generational effect of overweight and obesity on educational outcomes. The youngest cohort was more susceptible to the negative consequences of childhood overweight and obesity, but the oldest cohorts were not. This needs to be explored in further research.