BACKGROUND A number of studies have shown that light or moderate alcohol intake is associated with a reduced incidence of coronary heart disease (CHD) compared with non-drinkers. There is controversy as to whether this is due to a specific protective effect of alcohol or whether the increased risk of CHD in non-drinkers is due to selective migration of high-risk subjects from drinking into non-drinking categories. METHODS A prospective study of men aged 40-59 years drawn at random from one general practice in each of 24 British towns, and followed for cardiovascular morbidity and all-cause mortality over a 9.5-year period. RESULTS There was a shallow U-shaped relationship between alcohol intake and all major CHD events and a strong inverse association with fatal CHD events. In men with no evidence of CHD at screening, despite a 28% (non-significant) reduction in the relative risk of fatal CHD events in men drinking 2-6 drinks/day, the difference in absolute rates of CHD between the drinking categories was small. Men with symptomatic CHD showed a strong inverse association between alcohol and all major CHD events and fatal CHD events. In men without a doctor diagnosis of cardiovascular disorder, those drinking 2-6 drinks/day showed a 34% (non-significant) reduction in relative risk of CHD death compared to occasional drinkers but little reduction in total cardiovascular mortality and no reduction in all-cause mortality. CONCLUSION These findings provide weak support for the protective effect of alcohol on CHD and no encouragement for regular moderate drinking on the grounds of benefit to health.