Heavy maternal alcohol consumption during pregnancy has been associated with fetal growth retardation, but whether more moderate consumption also is associated remains a controversial issue. We examined moderate consumption in 1233 women with singleton livebirths, by calculating a weighted average of weekly consumption in the first 20 weeks from questions that asked about alcohol consumption before pregnancy, as well as any changes during the first half of pregnancy. The adjusted odds ratio for ("moderate") consumption of three or more drinks per week for low birthweight was 2.6 [95% confidence limits (CL) = 1.2, 5.7], and that for intrauterine growth retardation was 2.3 (95% CL = 1.2, 4.6). Examining the joint effect of smoking with alcohol consumption revealed associations that differed by outcome; we found a synergistic effect for low birthweight, but not for intrauterine growth retardation. Moderate alcohol consumers had an average birthweight decrement of 143 gm, which varied by smoking. There was little association of alcohol consumption with preterm delivery (< 37 weeks). Paternal alcohol consumption was not associated with any of the fetal growth measures after adjustment for other variables.