Moderate alcohol consumption is generally associated with reduced risk of type 2 diabetes. However, this beneficial effects of alcohol intake remains controversial due to inconsistent results across studies. The analysis was performed using data from the Ansung-Ansan cohort study. We categorized the participants into four groups—based on the baseline (one-point measure; non-drinking, <5 g/day, ≥5, <30 g/day, and ≥30 g/day) and follow-up (consumption pattern; never-drinking, light, moderate, and heavy drinking) measurement. At baseline, ≥30 g/day alcohol consumption increased the risk of incident diabetes (HR: 1.42; 95% CI, 1.10–1.85), but ≥5, <30 g/day alcohol consumption had no effects on the incident diabetes. Meanwhile, when using the alcohol consumption pattern, a heavy-drinking pattern increased the risk of incident diabetes (HR = 1.32, 1.01–1.73), but the light and moderate consumption pattern was associated with a reduced risk of type 2 diabetes (HR: 0.66; 0.50–0.87 and HR: 0.74; 0.57–0.95, respectively). At the end point of follow-up, the insulinogenic index (IGI), but not the insulin sensitivity index (ISI), differed among the groups. Alcohol consumption pattern had a J-shaped association with the incident type 2 diabetes in Korean men. The IGI showed an inverted J-shaped association according to alcohol drinking pattern, but the ISI was not a J-shape.