To determine whether the beneficial effects of alcohol on lipid concentrations are mediated by insulin levels, we performed a cross-sectional analysis in 2103 nondiabetic men and women aged 40 to 79 years from a general Japanese population in Hisayama. The multivariate-adjusted sum of fasting and 2-hour postloading insulin levels and the insulin resistance index significantly decreased with elevating alcohol intake levels in men (P < 0.01 for the trend) but not in women. No dose-response relations between alcohol intake and glucose levels were observed. In both sexes, high-density lipoprotein cholesterol (HDLC) significantly increased with elevated alcohol intake (P < 0.01), whereas total cholesterol and low-density lipoprotein cholesterol (LDLC) were inversely correlated with alcohol intake (P < 0.01). In contrast, triglycerides (TGs) levels in men showed a J-shaped relation to alcohol dose, with moderate drinkers (10-29 g/d) having the lowest levels. Estimates using regression models indicated that for men, 10% of the alcohol-induced increase in HDLC and 2% of the alcohol-induced decrease in LDLC were insulin mediated. It was also estimated for male subjects that 36% of the reduction in TGs due to low to moderate alcohol intake was mediated by low levels of insulin and that this insulin-mediated pathway reduced the positive alcohol-TG relation by 13% in cases of moderate to heavy drinking. Our data suggest that regular alcohol consumption dose-dependently increased insulin sensitivity among male nondiabetics, but the insulin-mediated beneficial effects of alcohol on lipid concentrations were relatively small.