Diaziquone (AZQ) is a lipophilic alkylating agent which crosses the blood-brain barrier and has marked antitumor activity in a broad spectrum of murine tumor systems. Myelosuppression has been the dose-limiting toxicity in phase I trials. In this study 36 patients with head and neck cancer received diaziquone. Thirty-one of these patients (28 male, 3 female) were evaluable for efficacy. The initial starting dose was 7 mg/m2/day X 5 days i.v. repeated every 28 days. Because of the severe myelosuppression encountered in the first four patients, the starting dose was decreased by 20% to 5.5 mg/m2/day X 5 days repeated every 28 days. The majority of patients were considered to be good-risk patients as evidenced by performance status (80% 0-1 Zubrod) and prior therapy. Even with this dosage reduction, myelosuppression (especially thrombocytopenia) was again the dose-limiting toxicity with 25% of patients experiencing granulocyte and platelet nadirs below 1000/mm3 and 50,000/mm3 respectively. Thirty-five percent of patients required a subsequent dosage reduction of 20% prior to receiving a second course of therapy. There was one complete (CR), four partial (PR) and three minor (MR) responses. All but the CR were of relatively short duration (mean of 30 days). The patient with a CR has remained disease-free for nearly 3 years. In this group of patients the activity of diaziquone as a single agent at this dose and schedule (CR + PR + MR = 26%; CR + PR = 16%) is less than that of methotrexate, bleomycin, and cis-platinum but is encouraging. Further trials utilizing combinations are warranted.