We conducted a retrospective case-control study to investigate a possible association between alcohol intake and stroke. Reported recent alcohol consumption and biochemical and hematologic markers of alcohol intake were examined for 230 patients with stroke (20 to 70 years old) and compared with concurrently collected data on controls matched for age, sex, and race. A single estimate of current intake was used as a measure of alcohol consumption. Among men, the relative risk of stroke (adjusted for hypertension, cigarette smoking, and medication) was lower in light drinkers (those consuming 10 to 90 g of alcohol weekly) than in nondrinkers (relative risk, 0.5), but was four times higher in heavy drinkers (consuming greater than or equal to 300 g weekly) than in nondrinkers. Because very few women in our study drank heavily, we were unable to determine whether heavy alcohol intake influenced the risk of stroke in women. With increasing serum concentrations of the biochemical markers of alcohol intake (aspartate aminotransferase, uric acid, and gamma-glutamyl transferase), we observed similar trends in the relative risk of stroke. Only the erythrocyte mean cell volume did not follow this pattern. We conclude that heavy alcohol consumption is an important and underrecognized independent risk factor for stroke in men, but our data are not adequate to settle the issue for women. Our conclusions are qualified by our reliance on reported recent alcohol consumption as the primary measure of intake.