Two cohorts of male bank employees aged 40-59 at baseline and free of coronary heart disease were pooled and followed during 25 years for all-cause and coronary mortality. The vital status was known for all 1,227 subjects, but in both banks, 8% of all deaths occurred without any clue to a possible cause. All-cause and coronary mortality were 36.9 and 10.5 per 100 subjects, respectively. In bivariate analysis (adjusting for age), systolic blood pressure, diastolic blood pressure, serum cholesterol and cigarette smoking predicted all-cause mortality. The same variables predicted coronary mortality. Body mass index was neither a predictor of all-cause nor of coronary mortality. In multivariate analysis, age, systolic blood pressure and cigarette smoking predicted independently all-cause mortality whereas the same variables and serum cholesterol predicted coronary mortality. For coronary mortality, the odds ratio (OR) of systolic blood pressure ≧ 160 against < 140 mm Hg was 2.68 (95% confidence interval, CI: 1.41-5.08). The OR for subjects with baseline serum cholesterol ≧260 versus < 260 mg/dl was 1.68 (95% CI: 1.31-2.51). The OR for cigarette smokers versus never smokers was 2.54 (95% CI: 1.39-4.64). Age-adjusted relative risk for all-cause mortality for subjects in the highest versus the lowest quintile of a multiple logistic function was 1.72 whereas for coronary mortality the relative risk was 4.94. These results confirm the long-term predictive power of the three major modifiable coronary risk factors when measured once in middle-aged male subjects.