The objective of the present work was to identify independent life-style factors for stroke and to estimate their causal contribution. The study is based on a random sample of the Copenhagen population selected in 1976 and stratified by age. The present analysis includes 12,961 subjects examined initially, aged 35 or over and without a previous cerebrovascular event, for whom information about life-style factors was recorded between 1976 and 1978. The outcome was the first-in-life stroke or transient ischemic attack during 12 years of follow-up. The events were ascertained at a second examination 5 years later and from hospital records and death certificates through 1988. Cox's regression model was used to estimate the effect on stroke risk of the factors recorded. In the period 1976-1988, 693 initial events were recorded in eligible responders. Among the life-style factors analyzed, a significant, independent effect was found for cigarette smoking, daily consumption of sleeping pills or tranquilizers and body mass index (BMI). There was a tendency for daily alcohol intake to be associated with lower risk, this could not be demonstrated for physical activity at leisure time. Among smokers, stroke risk was influenced by the number of cigarettes smoked, and daily alcohol intake was associated with a significantly lower risk. The effect of smoking decreased with age. BMI in smokers still had a significant effect on stroke risk but neither daily consumption of tranquilizers, nor physical inactivity at leisure time had a significant influence.