In a case--control study, we examined the relation between fetal loss and occupational exposure to antineoplastic drugs in nurses in 17 Finnish hospitals. The pregnancies studied occurred in 1973 through 1980 and were identified using three national sources: the Central Register of Health Care Personnel, the Hospital Discharge Registry, and policlinic data. Each nurse with fetal loss was matched with three nurses who gave birth. Data on health and exposure were obtained by self-administered, mailed questionnaires; a response rate of 87 per cent was achieved after three mailings. A statistically significant association was observed between fetal loss and occupational exposure to antineoplastic drugs during the first trimester of pregnancy: odds ratio = 2.30 (95 per cent confidence interval, 1.20 to 4.39). Analyses suggested associations between fetal loss and cyclophosphamide, doxorubicin, and vincristine, although the independent effect of each individual drug could not be specifically identified, since many nurses reported handling more than one of these agents. The results of this study, combined with existing data on animals and human beings, suggest that caution be exercised in the handling of these valuable drugs.