This study investigated the relation between body mass index (BMI) and the all-cause mortality rate among 7,985 European men. Starting around 1960, when all men were aged 40-59 years, mortality was followed for 15 years (1960-1975); starting around 1970, the survivors were followed for an additional 15 years (1970-1985). For the first and second follow-up periods, a BMI of 18.5-25 kg/m2 around 1960 and 1970, respectively, was considered the reference category. The authors found that the hazard ratios of mortality for a BMI of <18.5 kg/m2 was 2.1 (95% confidence interval (CI): 1.5, 2.8) for the first follow-up period and 1.7 (95% CI: 1.3, 2.2) for the second. A BMI of 25-30 kg/m2 was not related to increased mortality. Among never smokers, the hazard ratios for a BMI of >30 kg/m2 were 1.8 (95% CI: 1.2, 2.8) for the 1960-1975 follow-up period and 1.4 (95% CI: 1.0, 1.9) for the 1970-1985 follow-up period. A BMI of >30 kg/m2 was not related to increased mortality among current smokers. When mortality was followed for more than 15 years, the hazard ratio for a BMI of <18.5 kg/m2 declined and the hazard ratios for a BMI of >30 kg/m2 did not change. Underweight among those in all smoking categories and severe overweight in never smokers remained predictors of increased mortality when middle-aged men became older.