Risk factors for primary cerebral hemorrhage remain uncertain. The population-based Stroke Registry of Dijon provides data on the risk factors. Among residents of Dijon (France), 130 cases of primary cerebral hemorrhage hospitalized from 1985 to 1992 were matched with 130 controls by age and sex. The following data were collected: history of hypertension, alcohol consumption, tobacco consumption, history of coagulation disorder, diabetes mellitus, dyslipidemia, and infectious disease in the 7 days before admission. The following parameters were measured on admission: blood pressure, blood glucose, cholesterol, triglycerides, hematocrit, fibrinogen, prothrombin levels, platelet counts, prothrombin time, bilirubin, transaminases, γ-glutamyltransferase, and alkaline phosphatase. Electrocardiogram and Doppler ultrasound examination of cervical arteries were performed. Statistical analysis was performed by means of relative risk ratio for paired samples when dealing with proportions, and Student's t test for quantitative variables. A stepwise discriminant analysis was carried out to establish the relative weight of the different risk factors and their discriminant values. Among the qualitative data, the significant factors were history of hypertension, alcohol consumption, cardiac arrhythmia, atherosclerosis of carotid arteries and a previous infectious disease in the 7 days before admission. Among the quantitative data, the significant factors were early hypertension, high blood glucose levels, high hematocrit, and low cholesterol levels, in the acute stage of the stroke. After multifactorial analysis, only two factors were significant: hypertension and low cholesterol levels. Our population-based case-control study showed that hypertension and low cholesterol levels are the two discriminant risk factors for both lobar and basal ganglia primary cerebral hemorrhage. Therefore, treatment of hypercholesterolemia may increase risk of cerebral hemorrhage.