Alcohol consumption was assessed among 366 male high school students and 330 young men playing in sports teams. On the average, the athletes drank less wine, beer and spirits than the students and reported fewer drunkenness episodes. The association of alcohol drinking with sporting activity and with some independent variables was estimated by logistic regression techniques, fitting models for ordinal or nominal response variables. Point estimates of the odds ratios and their confidence limits showed: 1) an inverse relationship between alcohol drinking and sporting activity, especially as regards the consumption of spirits; 2) a strong positive relationship of alcohol drinking with both experimental and regular smoking; 3) peer alcohol drinking (best friend's and girlfriend's drinking) was the strongest predictor of the participant's alcohol consumption.