Elevated total leukocyte count in a biennial examination period is shown to be a significant (p = 0.001) predictor of cerebral infarction (CI) incidence in the subsequent 2 yr examination period, in a large Japanese cohort study. This association is not explainable on the basis of corresponding age, sex or blood pressure levels. The extent to which the association might be attributable to cigarette smoking habits could not be thoroughly examined with available data. Relative risks associated with a specific elevated leukocyte count may be larger among persons less than 65 yr of age than among older persons. When counts of specific leukocyte cell lines are considered a significant (p = 0.0006) role for neutrophil count emerges, while an additional predictive role for other leukocyte cell types could not be detected. In contrast, there is a suggestion that cerebral hemorrhage (CH) risk may be lower following an elevated leukocyte count. In particular, a negative association between lymphocyte count and CH incidence in the subsequent biennial examination cycle, is nearly significant (p = 0.07), in spite of a rather small number of CH cases in the sample.