Previous studies indicate that physical activity is related inversely to colon cancer risk. However, details regarding that association--whether a dose-response relation exists, whether the relation differs between non-obese and obese persons, the effect of long-term physical activity--are unclear. We examined these issues in the Physicians' Health Study (United States). Physical activity was assessed at baseline among 21,807 men, aged 40 to 84 years, and again 36 months later. Men were followed for an average of 10.9 years (from baseline) during which 217 developed colon cancer. After adjusting for potential confounders (including age, obesity, and alcohol intake), the relative risks for colon cancer associated with vigorous exercise in times per week (< 1, 1, 2-4, 5+, at baseline) were 1.0 (referent); 1.1 (95% confidence interval [CI] = 0.7-1.7); 1.2 (CI = 0.8-1.6); and 1.1 (CI = 0.7-1.6), respectively; P trend = 0.6. Physical activity was not associated significantly with colon cancer risk either among non-obese or obese men. When we used physical activity assessments at baseline as well as at 36 months, physical activity again was unrelated to colon cancer risk. These data do not support the hypothesis that physical activity reduces the risk of colon cancer. Plausible alternate explanations for the null finding include misclassification of physical activity and the potential for increased surveillance for colon cancer ('screening effect') among those physically active.