The relationship between alcohol consumption and angiographically documented coronary artery disease was examined among 484 men who underwent coronary arteriography. Severity of coronary artery disease was assessed by the number of arteries, or the number of segments, with at least one in which the luminal diameter is reduced by > or = 50%. Alcohol and tobacco consumption were assessed by a technician according to a standardized questionnaire, and an estimate of the average consumption level prior to the onset of the clinical symptoms was used for analysis. A decrease in the average coronary artery disease scores was observed with increasing alcohol consumption until a mean consumption of about 100 ml.day-1. After adjustment for age, town, clinical group and tobacco consumption, the same pattern of association was obtained. Similar trends were also present with alcohol consumption restricted to wine, beer and other alcoholic beverages. Taking into account classical risk factors of the disease and blood lipid levels at the time of angiography, did not modify these trends but less significant differences were observed. These findings provide additional evidence that the reduction of the clinical coronary risk observed in moderate alcohol consumers might be secondary, at least partly, to an anatomically less extensive disease. This, however, does not justify the recommendation of alcohol for prevention purposes.