Purpose This phase I trial was designed to (1) establish the dose of O6-benzylguanine (O6-BG) administered intravenously as a continuous infusion that suppresses O6-alkylguanine-DNA alkyltransferase (AGT) levels in brain tumors, (2) evaluate the safety of extending continuous-infusion O6-BG at the optimal dose with intracranially implanted carmustine wafers, and (3) measure the pharmacokinetics of O6-BG and its metabolite. Patients and Methods The first patient cohort (group A) received 120 mg/m2 of O6-BG over 1 hour followed by a continuous infusion for 2 days at escalating doses presurgery. Tumor samples were evaluated for AGT levels. The continuous-infusion dose that resulted in undetectable AGT levels in 11 or more of 14 patients was used in the second patient cohort. Group B received the optimal dose of O6-BG for 2, 4, 7, or 14 days after surgical implantation of the carmustine wafers. The study end point was dose-limiting toxicity (DLT). Results Thirty-eight patients were accrued. In group A, 12 of 13 patients had AGT activity levels of less than 10 fmol/mg protein with a continuous-infusion O6-BG dose of 30 mg/m2/d. Group B patients were enrolled onto 2-, 4-, 7-, and 14-day continuous-infusion cohorts. One DLT of grade 3 elevation in ALT was seen. Other non-DLTs included ataxia and headache. For up to 14 days, steady-state levels of O6-BG were 0.1 to 0.4 μmol/L, and levels for O6-benzyl-8-oxoguanine were 0.7 to 1.3 μmol/L. Conclusion Systemically administered O6-BG can be coadministered with intracranially implanted carmustine wafers, without added toxicity. Future trials are required to determine if the inhibition of tumor AGT levels results in increased efficacy.