The effect of alcohol consumption on breast cancer risk was investigated in a hospital-based, case-control study of 808 patients (349 cases and 459 controls) in Montpellier (France). Semi-quantified diet history including beverage consumption as well as relevant medical and personal characteristics were assessed by interview. A dose-rate relationship for total alcohol consumption was found with unadjusted odds ratios ranging from 1.8 (95% CI 1.2-2.8) for 1 to 2 drinks per week to 3.5 (95% CI 2.0-6.1) for more than 17 drinks per week, in comparison with less than 1 drink per week. Confounding by known or suspected determinants of breast cancer, smoking, use of oral contraceptives, lipid and vitamin consumption was looked into. Significant interactions were found with level of education and lipid consumption, with a higher risk for alcohol in women having had less schooling or consuming less fat. Adjustment with respect to the other risk factors did not modify the relationship. There was a significant risk increase both for wine and stronger drinks. Along with several other studies, our results give support to the hypothesis that alcohol is a risk factor in breast carcinogenesis.