Prescription frequencies in the first trimester were compared for miconazole, clotrimazole, nystatin, candicidin, aminacrine compounds, and metronidazole before deliveries involving congenital anomalies versus those not linked to congenital anomalies. Prescriptions before spontaneous abortions were compared with those in the first trimester of deliveries and with those before legal abortions. No statistically significant association was observed with any of these agents for the overall frequency of birth defects or for specific birth defects analyzed (cardiovascular defects, oral clefts, and spina bifida). Two hundred fifty miconazole exposures among 4264 spontaneous abortions, compared with 2236 in the first trimester of 55,736 deliveries, provided an estimated relative risk of 1.4 (95% confidence limits 1.2-1.5). One hundred twelve treatments with clotrimazole among the spontaneous abortions, compared with 1086 among the deliveries, provided a relative risk of 1.4 (95% confidence limits 1.1-1.6). In contrast, large numbers of exposures to nystatin and aminacrine compounds did not show this association, suggesting that spontaneous abortions are caused by the imidazole agents miconazole and clotrimazole rather than the condition being treated. Because many associations were examined without previous hypotheses, and because the data were inadequate to show an elevated risk for clotrimazole when comparing spontaneous with legal abortion exposures, these findings are considered to be a signal for further studies rather than definitive in themselves.