Lower cardiovascular mortality rates in lighter drinkers (versus abstainers or heavier drinkers) in population studies have been substantially due to lower coronary artery disease (CAD) mortality. Controversy about this U-shaped curve focuses on whether alcohol protects against CAD or, because of other traits, whether abstainers are at increased risk. Inclusion of ex-drinkers among abstainers in some studies has led to speculation that this might be the trait increasing the risk of abstainers. This new prospective study among 123,840 persons with 1,002 cardiovascular (600 CAD) deaths showed that ex-drinkers had higher cardiovascular and CAD mortality risks than lifelong abstainers in unadjusted analyses, but not in analyses adjusted for age, gender, race, body mass index, marital status and education. Use of alcohol was associated with higher risk of mortality from hypertension, hemorrhagic stroke and cardiomyopathy, but with lower risk from CAD, occlusive stroke and nonspecific cardiovascular syndromes. Subsets free of baseline cardiovascular or CAD risk had U-shaped alcohol-CAD curves similar to subsets with baseline risk. Among ex-drinkers, maximal past intake and reasons for quitting (medical versus non-medical) were unrelated to cardiovascular or CAD mortality. These data show that: (1) alcohol has disparate relations to cardiovascular conditions; (2) higher cardiovascular mortality rates among ex-drinkers are due to confounding traits related to past alcohol use; and (3) the U-shaped alcohol-CAD relation is not due to selective abstinence by persons at higher risk. The findings indirectly support a protective effect of lighter drinking against CAD.