Abstract The frequency of drinking alcohol among 616 women who aborted spontaneously (cases) was compared with that among 632 women who delivered after at least 28 weeks gestation (controls). 17·0% of cases reported drinking twice a week or more during pregnancy whereas among controls, only 8·1% of women reported drinking twice a week or more. The hypothesis that drinking during pregnancy is associated with spontaneous abortion was tested by maximum-likelihood logistic regression analysis. The adjusted-odds ratio for this association was 2·62. We estimate that more than one-quarter of pregnant women drinking twice a week or more are likely to abort, compared with about 14% among women who drink less often. Consideration of wine, beer, and spirits separately suggested that the minimum harmful dosage was one ounce of absolute alcohol. Several potentially confounding variables, including maternal age, gestation, prior spontaneous abortions, smoking, and nausea/vomiting, were controlled in the analysis. The association between drinking during pregnancy and spontaneous abortion did not vary with these factors. Even moderate consumption of alcohol during pregnancy is a risk factor for, and may be a cause of, spontaneous abortion. Among the possible mechanisms, acute fetal poisoning seems the most likely, although chronic poisoning is also possible.