With progress in stroke prevention it is important to evaluate epidemiological trends over a long period and in an unselected population. We estimated changes in stroke incidence, the case-fatality rate, risk factors, and the use of primary prevention, based on a continuous registry of a well-defined population, from 1985 to 2004. We recorded 3142 ischemic strokes, 341 primary cerebral hemorrhages and 74 subarachnoid hemorrhages. During the 20-year study period, age at first stroke increased by 5 years in men and 8 years in women. Comparing the periods 1985-1989 and 2000-2004, the age - and sex-adjusted incidence rates of first-ever strokes were stable, except for lacunar strokes, the incidence of which increased significantly (p = 0.05), and also cardio-embolic stroke, the incidence of which fell significantly (p = 0.01). Twenty-eight-day case-fatality rates fell significantly, especially for lacunar stroke (p = 0.05) and primary cerebral hemorrhage (p = 0.03). The frequency of hypercholesterolemia and diabetes increased significantly (p < 0.01). In contrast, the frequency of myocardial infarction fell significantly (p = 0.02). The frequency of smoking and diastolic blood pressure J 90 mmHg also fell, but the difference was not significant because of missing data. The frequency of pre-stroke antiplatelet and anticoagulant treatment increased significantly (p < 0.01). The age - and sex-adjusted incidence rates of stroke in Dijon, France, have thus been stable for the past 20 years. Age at stroke onset has increased, the case-fatality rate has fallen, antiplatelet treatment is more frequent, and the frequency of some pre-stroke risk factors has fallen.