To examine the combined influence of alcohol use and comorbidity on 20-year mortality in older adults (average age 66 at the time of the baseline survey). DESIGN: Longitudinal analysis of a national probability sample-based cohort study. SETTING: Data sources were the National Health and Nutrition Examination Survey I (NHANES I), 1971-1974, and the NHANES Epidemiologic Followup Survey, 1992. PARTICIPANTS: Four thousand six hundred ninety-one adults aged 60 and older who provided data on alcohol use. MEASUREMENTS: The prevalence of at-risk drinking in older adults in the United States and the 20-year all-cause mortality risk associated with it. At-risk drinking status was determined from amount of alcohol consumed and comorbidities, using a previously validated method. RESULTS: The prevalence of at-risk drinking in the United States between 1971 and 1974 in older adults was 10% (18% of men, 5% of women). The majority of at-risk drinkers were identified as such because of their use of alcohol in amounts deemed risky in the presence of relevant comorbidities (69%) (e.g., drinking 2-3 drinks per day and having gout or anxiety or taking a medication for pain). In men, at-risk drinking was associated with higher mortality rates than not-at-risk drinking (hazard ratio=1.20, 95% confidence interval=1.01-1.41); abstinence was not associated with greater mortality. In women, neither abstinence nor at-risk drinking was associated with greater mortality rates. CONCLUSION: In this first, large population-based study of older adults examining the mortality risks of alcohol use and comorbidity, at-risk drinking was associated with greater mortality rates in men. These findings suggest that a lower threshold of alcohol use should be recommended for older adults with specific comorbidities to reduce mortality risks.