Summary We examined the association between self-reported occupational and recreational physical activity and the subsequent risk of colorectal cancer in a population-based cohort in Norway. During a mean follow-up time of 16.3 years for males and 15.5 years for females, 236 and 99 colon cancers and 170 and 58 rectal cancers were observed in males and females, respectively, among 53 242 males and 28 274 females who attended the screening between 1972 and 1978. Physical activity at a level equivalent to walking or bicycling for at least four hours a week during leisure-time was associated with decreased risk of colon cancer among females when compared with the sedentary group (RR=0.62, 95% CI 0.40-0.97). Reduced risk of colon cancer was particularly marked in the proximal colon (RR=0.51, 95% CI 0.28-0.93). This effect was not observed for occupational physical activity alone, probably due to a narrow range of self-reported physical activity at work among females. However, by combining occupational and recreational physical activity we observed an inverse dose-response effect as increasing total activity significantly reduced colon cancer risk (P for trend=0.04). Among males 45 years or older at entry to the study, an inverse dose-response effect was observed between total physical activity and colon cancer risk (P for trend = 0.04). We also found in males a stronger preventive effect for physical activity in the proximal as compared to distal colon. In addition, we found a borderline significant decrease in colon cancer risk for occupational physical activity in males 45 years or older when compared to the sedentary group (RR = 0.74, 95% CI 0.53-1.04). All results were adjusted for age, body mass index, serum cholesterol and geographic region. No association between physical activity and rectal cancer was observed in males or females. The protective effect of physical activity on colon cancer risk is discussed in regard to energy balance, dietary factors, age, social class, body mass index and gastrointestinal transit time.Keywords: physical activity; colorectal cancer; cohort study; gender differences; subsites Cancer of the large intestine is one of the most common neoplasms in western countries (Muir et al., 1987; Engeland et al., 1993). Recently, the role of exercise in the aetiology of colon carcinogenesis has drawn particular interest. A growing number of epidemiological studies have reported a protective effect of occupational physical activity on colon cancer risk (Garabrant et al., 1984;Gerhardsson et al., 1986;Brownson et al., 1989;Peters et al., 1989;Arbman et al., 1993; Chow et al., 1993;Fraser and Pearce, 1993). Others have observed that recreational physical activity protects against colon cancer (Wu et al., 1987;Slattery et al., 1988;Gerhardsson et al., 1988;Severson et al., 1989;Ballard-Barbash et al., 1990;Lee et al., 1991;Markowitz et al., 1992; Giovannuci et al., 1995). In contrast, the association between physical activity and risk of rectal cancer is more inconsistent (Vena et al., 1985;Ge...