The relation between the level of total serum cholesterol and stroke is controversial. The relation between serum total cholesterol and subtypes of stroke was examined in the participants of a community-based mass screening program in Okinawa, Japan. A total of 38,053 subjects, whose serum level of cholesterol had been determined during a mass screening carried out in 1983, were examined to see whether they had experienced stroke during a 3-year period from 1988 to 1991. Of them, 315 subjects aged 33-93 years (174 men, 141 women) had had a stroke during that period. The types of stroke were cerebral infarction in 164, cerebral hemorrhage in 111, subarachnoid hemorrhage in 19, and others in 21. In men, the odds ratio of cerebral hemorrhage was 0.71 (95% confidence interval, 0.55-0.95), and the odds ratio of cerebral hemorrhage associated with serum level of cholesterol ≤167 mg/dl, 168-191 mg/dl, 192-217 mg/dl, and ≥218 mg/dl were 1.00 (reference), 0.70 (0.38-1.30), 0.77 (0.55-1.08), 0.73 (0.56-0.96), respectively. Lower serum cholesterol was an independent predictor of cerebral hemorrhage in men. (Jpn Circ J 1999; 63: 53 - 58)