Several investigators have observed an association between alcohol consumption and elevated glucose levels, raising the possibility that alcohol may increase the risk of diabetes. This hypothesis was evaluated prospectively among 85,051 women participating in the Nurses' Health Study who were 34 to 59 years of age in 1980 and had no history of cancer, coronary heart disease, or diabetes. At baseline, participants completed an independently validated dietary questionnaire which included information on the consumption of beer, wine, and liquor. Incident cases of non-insulin-dependent diabetes were reported on follow-up questionnaires sent in 1982 and 1984 (98% response to at least one follow-up); 526 cases were confirmed by a supplementary questionnaire regarding symptoms, laboratory values, and treatment. The risk of diabetes decreased monotonically with increasing alcohol consumption (chi trend = -9.4, p less than 0.0001). Compared with nondrinkers, women consuming 5-14.9 g of alcohol per day (about 4-10 drinks per week) had an age-adjusted relative risk of diabetes of 0.4 (95% confidence interval (CI) 0.3-0.6); for 15 g or more per day, the relative risk was 0.3 (95% CI 0.2-0.4). However, a strong inverse association between alcohol drinking and body weight explained much of the apparent protective effect of alcohol. After simultaneous adjustment for Quetelet index (weight (kg)/height (m)2), family history of diabetes, total caloric intake, and age, the relative risk of diabetes for consumers of 5-14.9 g per day was 0.8 (95% CI 0.6-1.2), and for women who drank 15+ g per day, the relative risk was 0.6 (95% CI 0.3-0.9). These data provide no support for the hypothesis that moderate alcohol intake increases the risk of non-insulin-dependent diabetes.