A population-based case-referent study on physical activity (during working and recreational hours) and colon cancer was performed in Stockholm in 1986-1988. The study included 1,081 subjects. Low physical activity was associated with an excess risk of colon (but not rectum) cancer for both men and women, showing a dose-response relationship with decreasing levels of physical activity. The effect was seen in the left colon (relative risk = 3.2, 95% confidence interval = 1.5-7.0) rather than in the right colon (relative risk = 1.1, 0.5-2.5). These results persisted after adjustment for year of birth, gender, body mass, intake of total energy, protein, total fat, dietary fibre, and browned meat surface.