The Copenhagen Male Study is a prospective, cardiovascular cohort study initiated in 1970 and consisting of 5249 employed men aged 40-59 years. A total of 4710 men, who had reported their smoking habits and were free of ischaemic heart disease, had their mortality recorded over a 17-year period: 585 men suffered a first incident of ischaemic heart disease (IHD), and 248 cases were fatal. There was a strong social gradient in the risk of IHD (Kendall's Tau B = 0.12, P less than 0.001). Adjusting for age, blood pressure, physical activity, body mass index and alcohol consumption in a multiple logistic regression equation, men in the lowest social class had a relative risk (95% confidence interval) of IHD of 3.6 (2.5-5.3) compared to men in the highest social class. We determined whether differences in smoking habits could explain at least some of this large increase in risk. Adjustment for the above factors and also inclusion of the form of tobacco smoked, the amount of tobacco smoked and presence or absence of inhalation, had very little effect on the estimate: the relative risk was 3.5 (2.4-5.2). There was no social gradient in age at the start of smoking. According to smoking habits, comparing social class V with social class I, the relative risk was 7.7 (2.6-22.4) in cigarette smokers, 6.0 (1.1-32.1) in pipe smokers, 3.5 (1.7-7.1) in mixed smokers, 2.25 (0.4-12.9) in cheroot smokers, 3.8 (2.4-5.9) in all smokers, 1.95 (0.8-4.6) in ex-smokers, and 4.7 (1.01-22.2) in non-smokers. In the upper social classes, 50-75% of IHD events could be ascribed to smoking, and in the lowest classes only about 20%. We conclude that the substantial social inequalities in risk of ischaemic heart disease are not accounted for by differences in smoking habits.