The CNS Cancer Consortium has conducted a phase III study comparing diaziquone (AZQ) with carmustine (BCNU) in the treatment of adults with primary anaplastic glial brain tumors. Patients eligible for this study were 18 years of age or older at the time of biopsy, subtotal resection, or gross total resection of an anaplastic glial brain tumor. Within 3 weeks of surgery, patients received whole brain radiotherapy at 1.7 to 2 Gy per fraction to a total whole brain dose of 42-48 Gy. This was followed by a boost to the tumor bed as ascertained by computed tomography (CT), angiography, and/or magnetic resonance imaging (MRI) of 1.7 to 2 Gy per fraction to a dose of 12-19 Gy. The recommended cumulative dose to the tumor bed was therefore 55-61 Gy. At 8 weeks following radiotherapy, patients were randomized to receive either AZQ at 15 mg/day for 3 days i.v. every 4 weeks or BCNU at 200 mg i.v. every 8 weeks. Chemotherapy was continued for at least 1 year unless death occurred, treatment failure was declared, or toxicity necessitated alteration of therapy. In the 249 randomized patients, there was no difference between the AZQ- and BCNU-treated patients in age, sex distribution, race, tumor histology, type of surgical resection, or Karnofsky performance status (KPS). Age and KPS at the initiation of therapy and tumor histology were the best overall predictors of survival. The type of chemotherapy (AZQ vs BCNU) was not predictive of survival. Two-year Kaplan-Meier survival was 22% in the AZQ-treated patients and 25% in BCNU-treated patients. In an analysis of radiotherapy administered we found that, within the range of doses required for this study, there was no influence of whole brain dose, boost dose, total dose, or size of the boost field on survival. The institution providing radiotherapy (teaching hospital vs nonteaching facility) did not influence survival.