In a prospective study of risk factors for ischaemic heart disease 792 54 year old men selected by year of birth (1913) and residence in Gothenburg agreed to attend for questioning and a battery of anthropometric and other measurements in 1967. Thirteen years later these baseline findings were reviewed in relation to the numbers of men who had subsequently suffered a stroke, ischaemic heart disease, or death from all causes. Neither quintiles nor deciles of initial indices of obesity (body mass index, sum of three skinfold thickness measurements, waist or hip circumference) showed a significant correlation with any of the three end points studied. Statistically significant associations were, however, found between the waist to hip circumference ratio and the occurrence of stroke (p = 0.002) and ischaemic heart disease (p = 0.04). When the confounding effect of body mass index or the sum of three skinfold thicknesses was accounted for the waist to hip circumference ratio was significantly associated with all three end points. This ratio, however, was not an independent long term predictor of these end points when smoking, systolic blood pressure, and serum cholesterol concentration were taken into account. These results indicate that in middle aged men the distribution of fat deposits may be a better predictor of cardiovascular disease and death than the degree of adiposity.