We investigated several aspects of the role of physical activity in colon and rectal cancer etiology that remain unclear in the European Prospective Investigation into Nutrition and Cancer. This cohort of 413,044 men and women had 1,094 cases of colon and 599 cases of rectal cancer diagnosed during an average of 6.4 years of follow-up. We analyzed baseline data on occupational, household, and recreational activity to examine associations by type of activity, tumor subsite, body mass index (BMI), and energy intake. The multivariate hazard ratio for colon cancer was 0.78 [95% confidence interval (95% CI), 0.59-1.03] among the most active participants when compared with the inactive, with evidence of a dose-response effect (P trend = 0.04). For right-sided colon tumors, the risk was 0.65 (95% CI, 0.43-1.00) in the highest quartile of activity with evidence of a linear trend (P trend = 0.004). Active participants with a BMI under 25 had a risk of 0.63 (95% CI, 0.39-1.01) for colon cancer compared with the inactive. Finally, an interaction between BMI and activity (P interaction = 0.03) was observed for right-sided colon cancers; among moderately active and active participants with a BMI under 25, a risk of 0.38 (95% CI, 0.21-0.68) was found as compared with inactive participants with BMI >30. No comparable decreased risks were observed for rectal cancer for any type of physical activity for any subgroup analyses or interactions considered. We found that physical activity reduced colon cancer risk, specifically for right-sided tumors and for lean participants, but not rectal cancer. (Cancer Epidemiol Biomarkers Prev 2006;15(12):2398 -407)