Thirty-one non-hemophilic patients with acquired antibodies to factor VIII were entered into a prospective, randomized, multi-institutional trial to determine the safety and efficacy of prednisone (P), cyclophosphamide (C), or the combination in the treatment of this disorder. Patients were eligible if they had antibody demonstrated by the Bethesda assay, had not received these agents previously, and gave informed consent. All patients were treated initially with P, 1 mg/kg, for 3 weeks. If the antibody persisted and there was no rise in factor VIII activity, patients were randomized to either continue P for an additional 6 weeks; taper P and begin C, 2 mg per kg; or continue P and add C. Antibody disappeared in 10 during the initial P therapy, and in three of four others randomized to continue on P; one patient was discontinued because of an herpetic infection. Antibody disappeared in three of six patients treated with C alone, and five of ten given C and P. The titer of antibody was significantly lower in responders than in non-responders (p = 0.003), but seven patients with titers of more than five Bethesda units had complete remissions. There was no difference in antibody titers between those responding to P and those responding to C. We conclude that all patients with acquired antibodies to factor VIII should receive initial management with P, and that C is effective as second-line therapy for many of those who are steroid-resistant.