It is inconclusive whether moderate alcohol consumption reduces the diabetes risk. We observed the development of impaired fasting glucose or type 2 diabetes according to the amount of alcohol intake and body mass index. The annual health evaluation data of 2,500 male workers from 2002 to 2006 were reviewed retrospectively deleting personal identification code. The information contained sex, age, medical history, smoking status, alcohol consumption, participating regular exercise, anthropometric, and biochemistry measurement. Impaired fasting glucose or diabetes was determined when fasting plasma glucose was > or =100mg/dL. Thousand seven hundred seven subjects were eligible after excluding medical history of diabetes or fasting glucose > or =100mg/dL at baseline. The relative risks of its development in group of taking 1-14, 15-29, and > or =30.0g ethanol were 0.842 (95% confidence interval [CI], 0.603-1.176), 1.068 (95% CI, 0.736-1.551), and 1.019 (95% CI, 0.662-1.568) within normal weight group, 1.164 (95% CI, 0.795-1.705), 1.421 (95% CI, 0.947-2.133), and 1.604 (95% CI, 1.031-2.495) within overweight group, and 1.498 (95% CI, 1.042-2.153), 1.634 (95% CI, 1.091-2.447), and 1.563 (95% CI, 1.019-2.396) within obese group each after adjusting age, family history of diabetes, smoking, exercise, serum fasting glucose, aspartate aminotransferase, and gamma-glutamyltransferase with nondrinkers as a reference group. Not only high alcohol consumption but also moderate drinking was related with higher incidence of impaired fasting glucose or diabetes in obese Korean men.