This cross-sectional analysis examined associations between alcohol intake and subjective health in a random sample (n = 6,040) drawn from the general population aged 25-64 years in Finland in 1992. Self-reported health was good for 3,375 persons and average or poor (suboptimal) for 2,665 persons. Crude odds ratios suggested a U-shaped pattern between alcohol intake and suboptimal health. The pattern took more of a J-shape after data were controlled for sex, age, education, marital status, lack of close friends, being on a disability pension, smoking, being an ex-drinker, and having decreased one's alcohol intake during the past 12 months because of health problems. An interaction was found between alcohol and smoking. The pattern of alcohol odds ratios showed a J-shaped association among never smokers, and a similar pattern was suggested among ex-smokers and current smokers. Among never smokers, the lowest risk was found at the alcohol consumption level of 100-199 g/week (odds ratio (OR) = 0.58, 95% confidence interval (CI) 0.38-0.89). The highest risk was found among persons who regularly smoked > or = 20 cigarettes per day and drank > or = 300 g/week (OR = 4.44, 95% CI 2.36-8.36). The risk for ex-drinkers did not differ from that for lifelong abstainers (OR = 0.89, 95% CI 0.62-1.28), but persons who had decreased their alcohol intake during the past 12 months because of health problems had a higher risk (OR = 1.21, 95% CI 1.05-1.39). The authors conclude that moderate alcohol intake is related to a self-perception of good health.