The alcohol-breast cancer hypothesis is important because (1) breast cancer is a major source of morbidity and mortality, (2) alcohol consumption is common, and (3) drinking is modifiable. Reports from more than 50 epidemiologic investigations of this hypothesis have now appeared. A recent metaanalysis of these studies indicates both a modest positive association between alcohol and breast cancer (an approximately 25% increase in risk with daily intake of the equivalent of two drinks) and a dose-response relation. Data suggest that risk increases with consumption of alcohol in general, regardless of beverage type. Several factors, including age, weight, and estrogen usage, have been shown to modify this relation in some studies. The authors discuss a series of methodologic issues in the study of alcohol and breast cancer. These include error in alcohol assessment, difficulties in evaluating small relative risks, and the potential for confounding. Several biologic mechanisms could account for an alcohol-breast cancer relation, with increasing attention being paid to a possible mediating effect of reproductive steroid hormones. Animal studies are a relatively recent development in this area; results have been mixed. Incorporation of more refined temporal, quantitative, and qualitative indicators of alcohol exposure in future epidemiologic studies would be valuable, as would further exploration of the endocrine and other metabolic effects of moderate alcohol consumption. The alcohol-breast cancer hypothesis remains intriguing, but causality has not been established.