From 1965 to 1968, the Honolulu Heart Program began following 8,006 men in a prospective study of cardiovascular disease. At the time of study enrollment, an estimate of current 24-hour habitual physical activity was collected from each subject. On the basis of a calculated physical activity index, subjects were classified as being inactive, partially active, or active. This report examines the relation between the levels of physical activity and stroke that occurred among 7,530 of the men over 22 years of follow-up. Risk of stroke was examined separately in younger (45–54 years) middle-aged men and older (55–68 years) middle-aged men. Among the older men, those who were inactive or partially active experienced a three- to fourfold excess incidence of hemorrhagic stroke as compared with active men (p < 0.01). There was a two- to threefold excess of intracerebral hemorrhage in men who were inactive or partially active as compared with those who were active (p < 0.05). An excess of subarachnoid hemorrhage was observed in inactive older men, with only one event occurring in those who were active (p < 0.05). After exclusion of subjects with hypertension, diabetes mellitus, and left ventricular hy pertrophy, the relative risk of hemorrhagic stroke for inactive men versus active men was 3.7(95% confidence interval (Cl) 1.3–10.4). In older men who did not smoke cigarettes, the relative risk of thromboembolic stroke among inactive men versus active men was 2.8 (95% Cl 1.2–6.7), and when partially active older men were compared with those who were active, the relative risk was 2.4 (95% Cl 1.0–5.7). These findings persisted after control for the residual effects of systolic blood pressure and other risk factors for stroke. Benefits of physical activity in reducing the risk of thromboembolic stroke were not observed in men who smoked cigarettes. The authors conclude that physical activity may be important in reducing the risk of stroke, particularly among nonsmoking men in older middle age.