Lipid accumulation product (LAP), defined as a product of waist circumference and triglycerides, has recently been proposed as a predictor of cardiovascular disease and diabetes mellitus. The purpose of this study was to determine whether and how LAP is associated with alcohol drinking. Subjects were 21,378 men aged 35-60 years and they were divided by alcohol intake into non-, light (<22 g ethanol/day), heavy (≥22 and <44 g ethanol/day) and very heavy (≥44 g ethanol/day) drinkers. Relationships between alcohol intake and LAP were analyzed by using multivariate analyses with adjustment for age, smoking and habitual exercise. Log-transformed LAP levels in light drinkers and very heavy drinkers were significantly (P < 0.01) lower and higher, respectively, than the level in non-drinkers, and the levels were comparable in non- and heavy drinkers (non-drinkers, 1.335 ± 0.005; light drinkers, 1.290 ± 0.009; heavy drinkers, 1.348 ± 0.005 and very heavy drinkers, 1.414 ± 0.006). The inverse association of alcohol intake with LAP was more prominent in smokers and subjects without regular exercise than in non-smokers and subjects with regular exercise, respectively, while the positive association of alcohol with LAP was more prominent in non-smokers than in smokers. Odds ratio for hyperglycemia of subjects with vs. subjects without high LAP was significantly higher than a reference level of 1.00, and this association was not different among the four alcohol groups. There is a J-shaped relationship between alcohol intake and LAP, which is confounded by smoking and habitual exercise.