The objective of this study was to investigate the association between alcohol consumption and the risk of total cancer, and to estimate the proportion of total cancer attributable to drinking habit in Japanese men. From June through August 1990, a total of 21 201 Japanese men completed a self-administered questionnaire on various health habits, including alcohol consumption. During 153 389 person-years of follow-up through December 1997, we identified a total of 882 cases of cancer. We used Cox proportional hazards regression to estimate the relative risk of total cancer according to categories of alcohol consumption. The risk for total cancer was significantly higher in ex-drinkers than never-drinkers. There was a dose-response relationship between the amount of alcohol consumed and the risk of total cancer among current drinkers: multivariate RRs in reference to never-drinkers (95% confidence intervals (CI)) were 1.1 (0.8-1.3), 1.3 (1.0-1.7), and 1.3 (1.1-1.7) in current drinkers who consumed less than 22.8 g, 22.8-45.5 g, 45.6 g or more alcohol per day, respectively (P for trend <0.001). Estimated 17.9% (95% CI 3.1-30.5) of total cancer risk was attributable to drinking habit. In our findings, approximately 20% of the total cancer cases in Japanese men may be prevented by alcohol control.