Background. Samples of railroad male employees aged 40-59 were examined from companies operating in the north-west quadrant of the US in the late 1950s (n = 2571), and in Rome and surroundings, Italy in the early 1960s(n = 768) in an international co-oprative study on cardlovascular diseases.Methods. A number of cardiovascular risk factors (age, blood pressure, serum cholesterol, body mass index, subscapular skinfold, physical activity, smoking habits) were measured at year 0 and year 5, while data on mortality and causes of death were collected for 25 years.Results. Coronary heart disease (CHD) death rates were always higher in the US than in Italy; after 25 years the excess was 51%. Cox proportional hazards models with CHD deaths as endpoint and seven CHD risk factors as covarites showed only the coefficlents for age and smoking habits statistically different between the two cohorts. Changes in systolic blood pressure during the first 5-year follow-up were additional predictors of 5- to 25-year CHD mortality in both groups. Cholesterol plus age (greater in the US), subscapular skinfold (greater in the US) and body mass index (greater in Italy) 'explained' 67% of the observed differences in 25-year CHD mortality in the two cohorts.Conclusions. A large proportion of differential CHD mortality can be attributed to some traditional risk factors.