PURPOSE: The aim of this study was to assess the effects of smoking on the risk of colorectal cancer according to anatomic subsite. METHODS: Between 1979 and 1985 a population-based case-control study of cancer at multiple sites was performed in Montréal, which accrued over 4,000 males in total, including cases of colorectal cancer, other cancers, and population controls. The present analysis was restricted to the 585 cases with histologically proven adenocarcinoma of the large bowel, aged 35 to 70 years, who underwent face-to-face interviews and provided adequate smoking histories. Of these, 176 had cancer in the proximal colon, 179 had cancer in the distal colon, and 230 had rectal cancer. Our control group consisted of 405 cancer controls, whose tumor types were considered unrelated to smoking, and 500 population controls. RESULTS: We observed a positive association between cigar smoking and cancer of the rectum. We also found some suggestion of a weak positive association between cigarette smoking and cancer of the proximal colon, an inverse association with cancer of the distal colon, although neither was statistically significant, and no association with rectal cancer. CONCLUSIONS: Cigar smoking seems to be associated with the development of rectal cancer. If the positive association between cigarette smoking and cancer of the proximal colon is real, it might partially explain the proximal shift in the anatomic distribution of colorectal cancer that has been observed, because of the increasing prevalence of cigarette smoking during the middle of the 20th century.