Five cohorts of men ages 40-59 (Finland: 2 cohorts of 1,677 men; Netherlands: 1 cohort of 878 men; Italy: 2 cohorts of 1,712 men) were examined and evaluated for cardiovascular risk factors in 1959-1960 and subsequently followed-up for mortality over the next 25 years. Age-adjusted death rates from coronary heart disease were highest in Finland (244 per 1,000), intermediate in The Netherlands (195 per 1,000), and lowest in Italy (122 per 1,000) with a twofold range between the extremes. The Cox proportional hazards model was used for single cohorts and for the pools of national cohorts with coronary heart disease deaths as endpoints and 12 risk factors as covariates. It showed the significant and almost universal predictive value of these factors (with some rare exceptions). The most highly predictive values were age, blood pressure, total serum cholesterol, cigarette smoking, and physical activity (negative relationship). The prediction of events within each country using the risk function of the others produced errors ranging from -19% to +51%. The largest errors were those involving the Italian cohorts whose experience tended to underpredict coronary heart disease mortality elsewhere and to be overpredicted by the risk functions of the other countries. Solving a Cox model which included all the cohorts, and adding dummy variables for the identification of nationality, it appears that the relative risk, everything else being equal, is 1.49 and 1.34 for a Finnish man, compared with Italian and Dutch men, respectively.