This study was designed to test if the activity of a phase II agent, ifosfamide, would have been underestimated if it was tested exclusively in a population of children and young adults with recurrent osteosarcoma. The response rate to ifosfamide was compared in patients younger than 30 years of age with previously untreated osteosarcoma with metastases at diagnosis and/or unresectable primary tumors (stratum 1) with that of patients with recurrent osteosarcoma following adjuvant chemotherapy who were not previously exposed to ifosfamide (stratum 2). Evaluation of response was conducted 3 weeks after two courses of ifosfamide (2400 mg/m2 x 5 days) were administered 3 weeks apart. Nine of 33 (27%) evaluable patients in stratum 1 responded (1 complete and 8 partial responses) to ifosfamide. Among 30 evaluable patients in stratum 2, only 3 (10%) responded (1 complete and 2 partial responses; P = .04) Both groups of patients received equal doses of ifosfamide and experienced comparable toxicities. Results from this study suggest that the activity of new agents will be underestimated if tested in a population of heavily pretreated patients with recurrent disease. When possible, new chemotherapeutic agents should be tested in patients with a poor prognosis who have not been exposed to chemotherapy.