We conducted a cross-sectional study to clarify the dose-effect relationship of smoking habits with blood pressure in Japanese men. The subjects were 2781 normotensive male steelworkers ranging in age from 40 to 54 years. They were classified into five categories according to their smoking habits: non-smokers, exsmokers, light smokers, moderate smokers and heavy smokers. Age, body mass index, salt intake, physical activity, drinking habits, and levels of gamma-glutamyl transpeptidase, total serum cholesterol, uric acid, creatinine and plasma glucose were analysed as covariates. The association between smoking habits and blood pressure was evaluated using analysis of covariance. Our results showed that there was no significant differ-