As part of an on-going longitudinal study, 7895 men of Japanese ancestry living on the island of Oahu, aged 45-68 and free of evidence of prior stroke at entry examination, have been followed by re-examinations and surveillance. During ten years of follow-up 154 men developed thromboembolic stroke, 65 developed intracranial hemorrhage, and 19 developed stroke of unknown type. There were 79 deaths attributed to stroke. The independent risk factors for thrombo-embolic stroke were elevated blood pressure, glucose intolerance, age, electrocardiographic evidence of left ventricular hypertrophy or strain, cigarette smoking and proteinuria. Attributes associated with increased risk of intracranial hemorrhage were age, elevated blood pressure, cigarette smoking, serum uric acid and, inversely, serum cholesterol level. Electrocardiographic evidence of left ventricular hypertrophy or strain significantly increased the risk of cerebral hemorrhage, but was not associated with subarachnoid hemorrhage. In univariate analysis, there was an inverse relation between dietary fat intake and thrombo-embolic and total stroke incidence. An inverse relation was also shown between protein intake and total stroke incidence. These dietary relations became statistically not significant in multivariate analysis. No relation was found between salt intake and the incidence of stroke.