This Phase III trial tested the efficacy and safety of intra-arterial 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) for the treatment of newly resected malignant glioma, comparing intra-arterial BCNU and intravenous BCNU (200 mg/sq m every 8 weeks), each regimen without or with intravenous 5-fluorouracil (1 gm/sq m three times daily given 2 weeks after BCNU). All patients also received radiation therapy. A total of 505 patients were randomly assigned within the study. Fifty-seven patients were excluded, primarily because of neuropathology error, and the remaining 448 patients constituted the Valid Study Group. Of the total 505 patients, 190 patients could not receive intra-arterial BCNU and 315 patients were randomly assigned to receive intra-arterial (167 patients) and intravenous (148 patients) BCNU. Actuarial analysis (log-rank) demonstrated reduced survival for the intra-arterial group (p = 0.03). Serious toxicity was observed in the intra-arterial group; 16 patients (9.5%) developed irreversible encephalopathy with computerized tomography evidence of cerebral edema, and 26 patients (15.5%) developed visual loss ipsilateral to the infused carotid artery. Administration of 5-fluorouracil did not influence survival. The survival rate between the intravenous and the intra-arterial BCNU patients with glioblastoma multiforme did not differ, but was worse for intra-arterial BCNU patients with anaplastic astrocytoma than for those receiving intravenous BCNU (p = 0.002). Neuropathologically, intra-arterial BCNU produced white matter necrosis. It is concluded that intra-arterial BCNU is neither safe nor effective in prolonging survival when administered by the methods used in this study of newly diagnosed patients with malignant glioma.