Using data from a 12-year prospective study, we determined the importance of the pattern of alcohol consumption as a risk factor for type 2 diabetes in a cohort of 46,892 U.S. male health professionals who completed biennial postal questionnaires. Overall, 1,571 new cases of type 2 diabetes were documented. Compared with zero alcohol consumption, consumption of 15-29 g/day of alcohol was associated with a 36% lower risk of diabetes (RR = 0.64; 95% CI 0.53-0.77). This inverse association between moderate consumption and diabetes remained if light drinkers rather than abstainers were used as the reference group (RR = 0.60, CI 0.50-0.73). There were few heavy drinkers, but the inverse association persisted to those drinking >/=50 g/day of alcohol (RR = 0.60, CI 0.43-0.84). Frequency of consumption was inversely associated with diabetes. Consumption of alcohol on at least 5 days/week provided the greatest protection, even when less than one drink per drinking day was consumed (RR = 0.48, CI 0.27-0.86). Compared with infrequent drinkers, for each additional day per week that alcohol was consumed, risk was reduced by 7% (95% CI 3-10%) after controlling for average daily consumption. There were similar and independent inverse associations for beer, liquor, and white wine. Our findings suggested that frequent alcohol consumption conveys the greatest protection against type 2 diabetes, even if the level of consumption per drinking day is low. Beverage choice did not alter risk.