Ifosfamide (5 g/m2) was compared with its parent analogue cyclophosphamide (1.5 g/m2) in a randomized phase II study. Both drugs were given by 24-h intravenous (i.v.) infusion every 3 weeks along with i.v. bolus infusions of mesna (400 mg/m2), which was given every 4 h for nine administrations. Eligibility criteria included an age of 15-75 years, biopsy-proven advanced metastatic soft-tissue sarcoma, and a World Health Organization performance status of 0-2. Exclusion criteria were prior treatment with classic alkylating agents, a creatinine level of > 150 mumol/l, a bilirubin level of > 20 mumol/l, a leukocyte cell count of < 3.5 x 10(9)/l, and a platelet count of < 100 x 10(9)/l. A total of 171 patients were entered, 24 of whom were ineligible and 12, inevaluable, leaving 135 patients evaluable. In all, 67 patients were treated with cyclophosphamide, and the overall response rate was 7.5%. All responders to cyclophosphamide were patients who had not received prior chemotherapy (13% of 38 patients). Another 68 patients were given ifosfamide, 18% of whom responded to treatment. Of the 28 ifosfamide-treated patients who had received prior chemotherapy, 7% were responders. The response rate for the remaining 40 patients was 25%. The higher overall response rate (P = 0.13) obtained with less myelosuppression in ifosfamide-treated patients suggests that this agent may have advantages over cyclophosphamide in combination therapy.