Fifty-eight patients with high-grade astrocytoma were treated by members of the Childrens Cancer Study Group in a prospective randomized trial designed to study the effectiveness of chemotherapy as an adjuvant to standard surgical treatment and radiotherapy. Following surgical therapy, patients were assigned randomly to radiotherapy with or without chemotherapy consisting of chloroethyl-cyclohexyl nitrosourea, vincristine, and prednisone. Treatment with chemotherapy prolonged survival and event-free survival. Five-year event-free survival was 46% for patients in the radiotherapy and chemotherapy group, and 18% for patients in the radiotherapy-alone group. Five-year survival was similarly improved. The differences in outcome due to treatment were statistically significant after correcting for imbalances in important prognostic factors (event-free survival, p = 0.026; survival, p = 0.067). The presence of mitoses or necrosis in the tumor specimen was associated with poorer outcome. Patients whose initial surgery was limited to biopsy, and patients with basal ganglia lesions, also had significantly worse outcome. Chemotherapy administered at the time of recurrence in a small number of patients did not produce any long-term survivors. This study is to our knowledge the only randomized trial to investigate effectiveness of chemotherapy in the treatment of high-grade astrocytoma in children.