We conducted a randomized phase II study to determine the efficacy of dacarbazine (DTIC) in recurrent gliomas. Patients were randomly assigned to receive either DTIC 750 mg/m2 IV day 1 every 28 days (Arm A) or DTIC 200 mg/m2 IV days 1-5 every 28 days (Arm B). Pharmacokinetics were studied in 6 patients on each arm using HPLC analysis. Thirty-nine patients (30 male, 9 female), ages 27-67 years (median 53) were entered on the study (20 on Arm A, 19 on Arm B). No objective responses were seen. Median time to progression was 3 months. Median survival was 8 months. Treatment was generally well tolerated. Major toxicities were grade 1-2 nausea (33%). lethargy (28%), diarrhea (15%), alopecia (15%), and grade 3 neutropenia (8%). Four patients on Arm A had mild self-limited episodes of intravascular hemolysis occurring immediately after drug infusion, the mechanism of which is unknown. Mean AUC for DTIC, HMMTIC (5-[3-hydroxymethyl-3-methyl-1-triazeno] imidazole-4-carboxamide), and MTIC (5-[3-methyl-1-triazenol imidazole-4-carboxamide), in Arm A were 14.8, 0.17, and 1.15 mM min, respectively. Corresponding values for Arm B (on day 1 of 5) were 1.7, 0.06, and 0.29 mM min, respectively. The predicted HMMTIC and MTIC exposure over 5 days for Arm B, based on the day 1 data, is higher than with Arm A. We conclude that DTIC is well tolerated but does not have activity in patients with recurrent gliomas. The 5-day schedule appears less toxic, and pharmacokinetic studies show that it provides greater exposure to MTIC and HMMTIC compared to the one-day schedule.