The Copenhagen City Heart Study is a prospective study based on a randomly selected sample of an urban population of, initially, 19,698 participants followed since 1976. Risk factor analysis was based on the initial examination of 13,000 persons > or = 35 years old without previous stroke who responded to the first invitation. In the period 1976-1988, 696 initial cases were identified: 584 strokes, 106 transient ischemic attacks and 6 retinal-artery occlusions. We used the regression model of Cox based on a hierarchic system of risk factors that indicated the way they influence each other. This method distinguishes independent risk factors and estimates their causal influences on the risk of stroke. Among the basic variables analyzed in this paper, significant effects were found for age, sex, length of school education and income. There was a tendency for living alone to be a risk factor as opposed to living with someone, while no influence could be demonstrated for family history of stroke.