Background: Alcohol use has beneficial as well as adverse consequences on health, but few studies examined its role in the development of age-related frailty. Objectives: To describe the cross-sectional and longitudinal association between alcohol intake and frailty in older persons. Design: The Lausanne cohort 65+ population-based study, launched in 2004. Setting: Community. Participants: One thousand five hundred sixty-four persons aged 65-70 years. Measurements: Annual data collection included demographics, health and functional status, extended by a physical examination every 3 years. Alcohol use (AUDIT-C), and Fried's frailty criteria were measured at baseline and 3-year follow-up. Participants were categorized into robust (0 frailty criterion) and vulnerable (1+ criteria). Results: Few participants (13.0%) reported no alcohol consumption over the past year, 57.8% were light-to-moderate drinkers, while 29.3% drank above recommended thresholds (18.7% "at risk" and 10.5% "heavy" drinkers). At baseline, vulnerability was most frequent in non-drinkers (43.0%), least frequent in light-to-moderate drinkers (26.2%), and amounted to 31.9% in "heavy" drinkers showing a reverse J-curve pattern. In multivariate analysis, compared to light-to-moderate drinkers, non-drinkers had twice higher odds of prevalent (adjOR: 2.24; 95%CI:1.39-3.59; p=.001), as well as 3-year incident vulnerability (adjOR: 2.00; 95%CI:1.02-3.91; p=.043). No significant association was observed among "at risk" and "heavy" drinkers. Conclusion: Non-drinkers had two-times higher odds of prevalent and 3-year incident vulnerability, even after adjusting for their baseline poorer health status. Although residual confounding is still possible, these results likely reflect a healthy survival effect among drinkers while those who experienced health- or alcohol-related problems stopped drinking earlier.