Epidemiologic evidence suggests that alcohol consumption has distinctive associations with risk of ischemic and hemorrhagic strokes; these differences help explain apparent inconsistencies in the alcohol-stroke literature (Camargo CA Jr. Stroke 1989;20:1611-1626). To better define the impact of "moderate drinking" per se (i.e. usual consumption of < or = 2 drinks daily for men, and < or = 1 drink daily for women), the present author reviewed 26 case-control and cohort studies on this subject. There is substantial evidence that moderate drinking does not increase risk of ischemic stroke; studies remain divided, however, on the question of a "protective" association. Furthermore, although the evidence is not unanimous, two major cohort studies have found that even moderate drinking may increase risk of hemorrhagic stroke. Because ischemic strokes are 3-4 times more common than hemorrhagic strokes, the net impact of moderate drinking on stroke risk depends greatly on the relation between moderate drinking and ischemic stroke. It is most likely, however, that moderate drinking does not increase risk of all strokes combined.