We compared two chemotherapy regimens that included methotrexate (MTX), cisplatin (CDP), and doxorubicin (ADM) with or without ifosfamide (IFO) in patients with nonmetastatic osteosarcoma of the extremity. Patients age ≤ 40 years randomly received regimens with the same cumulative doses of drugs (ADM 420 mg/m(2), MTX 120 g/m(2), CDP 600 mg/m(2), and IFO 30 g/m(2)) but with different durations (arm A, 44 weeks; arm B, 34 weeks). IFO was given postoperatively when pathologic response to MTX-CDP-ADM was poor (arm A) or given in the primary phase of chemotherapy with MTX-CDP-ADM (arm B). End points of the study included pathologic response to preoperative chemotherapy, toxicity, and survival. Given the feasibility of accrual, the statistical plan only permitted detection of a 15% difference in 5-year overall survival (OS). From April 2001 to December 2006, 246 patients were enrolled. Two hundred thirty patients (94%) underwent limb salvage surgery (arm A, 92%; arm B, 96%; P = .5). Chemotherapy-induced necrosis was good in 45% of patients (48% in arm A, 42% in arm B; P = .3). Four patients died of treatment-related toxicity (arm A, n = 1; arm B, n = 3). A significantly higher incidence of hematologic toxicity was reported in arm B. With a median follow-up of 66 months (range, 1 to 104 months), 5-year OS and event-free survival (EFS) rates were not significantly different between arm A and arm B, with OS being 73% (95% CI, 65% to 81%) in arm A and 74% (95% CI, 66% to 82%) in arm B and EFS being 64% (95% CI, 56% to 73%) in arm A and 55% (95% CI, 46% to 64%) in arm B. IFO added to MTX, CDP, and ADM from the preoperative phase does not improve the good responder rate and increases hematologic toxicity. IFO should only be considered in patients who have a poor histologic response to MTX, CDP, and ADM.