In osteosarcoma, intraarterial (IA) administration of systemic treatment has been advocated to improve local tumor response preparing for, or even obviating, definitive surgery. Because data from the literature did not unequivocally support the local superiority of IA infusion, a comparative study was started in 1986. Preoperative chemotherapy consisted of 45 mg/m2 of doxorubicin on days 1 and 2; 12 g/m2 of high-dose methotrexate on days 15 and 22; and 3 g/m2 of ifosfamide on days 29, 30, 50, and 51 followed on days 31 and 52 by intravenous (IV) versus IA tourniquet infusion of cisplatin (DDP). A strict randomization of patients was not feasible. A balanced distribution of risk factors was strived for by stratifying and allocating the appropriate patients centrally. The infusion time was prolonged from 1 to 5 hours in the IV group, and the DDP dose was reduced from 150 to 120 mg/m2 in both arms when intolerable ototoxicity became apparent. A multivariate analysis was performed to exclude a bias on the response rates from risk factor distribution and from modifications of DDP infusion time and dosage. The overall fraction of histologic good responders (greater than 90% necrosis) was not found to be different after IA versus IV treatment (34/50 [68%] vs. 41/59 [69%]). Intraarterial instead of IV use of DDP within an aggressive systemic treatment does not seem to improve the local tumor response.