We evaluated the occurrence of acute nonlymphocytic leukemia (ANL) among 1399 women with ovarian cancer who were treated in five randomized clinical trials. Of the 1399 women, 998 had been treated with alkylating agents, and among these, 12 cases of ANL were observed; the expected number was 0.11. Ten patients with ANL had received melphalan, and two chlorambucil. ANL was not observed in 401 women who had been treated with surgery or radiation or both, without alkylating agents. The excess risk of ANL that was associated with alkylating-agent therapy was 5.8 cases per 1000 women per year, and the cumulative seven-year risk of ANL among patients who were treated with chemotherapy alone was indistinguishable from that observed in patients receiving both radiation and chemotherapy. A positive correlation between initial drug dose and the risk of ANL was suggested. These data underscore the need to assess other cytotoxic agents and regimens of drug administration to identify those that do not have harmful late effects.