Objective: to assess the relation between alcohol intake and mortality among seven cohorts of middle-aged and elderly Danes. Design: prospective population study with baseline assessment of alcohol- and tobacco consumption, educational level and body mass index, and a mean of 11.5 years follow-up of mortality. Subjects: 16 304 men and women aged 50 years or more. Main outcome measure: number and time of deaths from 1974 to 1995 as ascertained by the national central person register. Results: the effect of alcohol intake on mortality did not differ between middle-aged (50–64 years, mean = 56.6 years) and elderly subjects (>64 years old, mean = 69.9 years). There was a U-shaped risk function in both age groups, which persisted also when adjusting for age, sex, smoking habits, level of education and body mass index. Abstaining women had a relative risk of 1.29 (95% confidence limits 1.17–1.42) as compared with light drinkers (1–6 drinks per week), while the relative risk for abstaining men was 1.22 (95% confidence limits; 1.08 to 1.37) as compared with light drinkers. Heavy drinking women (>28 drinks per week) had a relative risk of 1.23 (95% confidence limits; 0.85 to 1.78) and heavy drinking men (more than 69 drinks per week) had a relative risk of 2.11 (95% confidence limits 1.66–2.69), both compared with light drinkers. Conclusion: among the middle-aged and elderly women and men, a light alcohol intake is associated with lower mortality than abstention or heavy drinking.