Malignant gliomas will affect 15,000-17,000 Americans each year and carry a dismal prognosis. Adjuvant chemotherapy is hampered by inadequate drug delivery, systemic toxicity, and a markedly variable biological sensitivity. Intraarterial (i.a.) therapy may enhance selectivity by improving tumor drug delivery and reducing systemic toxicity. Using melphalan given i.a., we studied the therapy of intracranial human glioma xenografts in male athymic nude rats (mean weight, 300 g) which were inoculated intracerebrally with D-54 MG and D-456 MG. On Days 6 and 7 (D-54 MG) or Days 9 and 10 (D-456 MG), rats randomized by body weight and treated with single-dose melphalan given i.a. at 0.5 or 0.75 mg produced significantly higher median survival (D-54 MG, Days 33 and 32; D-456 MG, Days 52 and 54, respectively) compared with i.a. saline (D-54 MG, Day 14, P < 0.001; D-456 MG, Day 24, P = 0.000) or melphalan given i.v. at 0.75 mg and 0.9 mg (D-54 MG only; Day 19, P < 0.001; Day 23, P < 0.001, respectively) and at 0.5 and 0.75 mg (D-456 MG only; Day 26 for both doses, P = 0.00). Although a dose-dependent increase in median survival (D-54 MG, 0.25 mg, Day 18; 0.5 mg, Day 28.5; 0.75 mg, Day 32.5) was observed with i.a. administered melphalan, no significant difference was apparent between 0.5 and 0.75 mg in either tumor model (D-54 MG, P = 0.15; D-456 MG, P = 0.37). Toxicity studies in nontumor-bearing athymic rats yielded a maximum tolerated dose of 0.8 mg for i.a. administered melphalan. This dosage was superior in spite of different xenograft permeabilities (apparent mean blood-to-tissue transport [K] values for alpha-aminoisobutyric acid, 5.8 for D-54 MG and 1.3 for D-456 MG). Pharmacokinetic experiments demonstrated a significant first pass advantage for i.a. (versus i.v.) melphalan. The short plasma half-life, marked antiglioma activity, and lack of requirement for metabolic activation indicate that i.a. melphalan holds considerable promise for human glioma therapy.