To perform a systematic review of studies reporting on the association between body mass index (BMI) and the risk of colorectal cancer, we conducted a meta-analysis and meta-regression analysis. The identified 56 studies were conducted among 7 213 335 individuals including 93 812 cases. Compared with BMI < 23.0 kg m(-2), BMI of 23.0-24.9, 25.0-27.4, 27.5-29.9 and > or = 30.0 kg m(-2) were associated with 14%, 19%, 24% and 41% increased risks, respectively. Asians and premenopausal women had sharply increased risk from BMI < 23 kg m(-2) to general 'normal' range (23-25 kg m(-2)). Each 5 kg m(-2) increment was associated with 18% increased risk. Meta-regression analysis indicated that the association was stronger for colon than rectal cancer (P < 0.001), for men than women (P < 0.001), for self-reported BMI than directly measured BMI (P < 0.001), and for studies adjusting for physical activity than not adjusting (P < 0.001). The variation of the reported risk estimates for the association can be partly explained by cancer site, sex, women menopausal status, BMI assessment and adjustment of confounding variables.