Objective: We have analysed the association between alcohol drinking before and during the three trimesters of pregnancy and risk of preterm birth of babies with normal weight for gestational age or with low weight for gestational age (SGA). Design: Case-control study. Setting: General and university hospitals in Italy. Subjects: Cases were 502 women who delivered preterm births o37 weeks gestation. The controls included 1966 women who gave birth at term (Z37 weeks of gestation) to healthy infants of normal weight (ie between 10th and 90th centile according to the Italian standard) on randomly selected days at the hospitals where cases had been identified. Interventions: Interview. Results: No increased risk of preterm birth was observed in women drinking one or two drinks/die in pregnancy, but three or more drinks/die increased the risk (multivariate odds ratios (OR) 2.0 for Z3 drinks during the first trimester, 1.8 during the second and 1.9 during the third). When the analysis was conducted separately for preterm births with normal weight or SGA, the increased risk was observed in preterm SGA only (multivariate OR for Z3 drinks/die during the first trimester ¼ 3.6, 95% confidence interval (CI) 1.3-11.1); the estimated multivariate OR for Z3 drinks/die during the first trimester of preterm babies with normal weight for gestational age was only slightly above unity and not statistically significant (multivariate OR 1.4, 95% CI 0.5-3.7). Conclusions: The study shows an increased risk in mothers who drink Z3 die units alcohol in pregnancy of preterm births. Sponsorship: Ministero della Ricerca Scientifica.