The purpose of this prospective study was to investigate the effect of alcohol consumption on the 10-year diabetes incidence.In 2001-2002, a random sample of 1514 men (18-89 years old) and 1528 women (18-87 years old) was selected to participate in the ATTICA study (Athens metropolitan area, Greece). Among various other characteristics, average daily alcohol intakes (abstention, low, moderate, high) and type of alcoholic drink were evaluated. Diabetes was defined according to American Diabetes Association criteria. During 2011-2012, the 10-year follow-up was performed.The 10-year incidence of diabetes was 13.4% in men and 12.4% in women. After making various adjustments, those who consumed up to 1 glass/day of alcohol had a 53% lower diabetes risk (RR=0.47; 95% CI: 0.26, 0.83) compared with abstainers, while trend analysis revealed a significant U-shaped relationship between quantity of alcohol drunk and diabetes incidence (P<0.001 for trend). Specific types of drinks were not associated with diabetes incidence; however, a one-unit increase in ratio of wine/beer/vodka vs. other spirits was associated with an 89% lower risk of diabetes (RR=0.11; 95% CI: 0.02, 0.67). The protective effect of low alcohol consumption on diabetes incidence was more prominent among individuals with stricter adherence to the Mediterranean diet (RR=0.08; 95% CI: 0.011, 0.70) and without the metabolic syndrome (RR=0.34; 95% CI: 0.16, 0.70).This work revealed the protective effect of modest alcohol consumption of particularly wine and beer against the long-term incidence of diabetes, possibly due to their pleiotropic health effects.