Although mercaptopurine is the thiopurine antimetabolite predominantly used in the treatment of childhood acute lymphoblastic leukemia (ALL), thioguanine (TG) is more potent than mercaptopurine in in vitro cytotoxicity studies in human leukemic cell lines and leukemic cells from patients with ALL. We conducted a pediatric Phase I trial of TG administered as a continuous i.v. infusion (CIV). A pharmacokinetically guided dose escalation was performed to define the dose rate of TG required to achieve a steady-state plasma concentration (Css) exceeding the target concentration of 1 microM, and then the maximum tolerated duration of infusion of TG at this dose rate was defined. Eighteen patients (median age, 18 years; range, 4-25 years) with refractory malignancies (16 solid tumors and 2 ALL) were enrolled in this study. The starting dose rate of 10 mg/m2/h administered for 24 h achieved an average Css of 0.9 microM (range, 0.7-1.2 microM). Therefore, the dose rate was escalated to 20 mg/m2/h, which achieved an average Css of 4.1 microM (range, 1. 0-8.3 microM). This disproportionate increase in the Css of TG suggested a capacity-limited (saturable) elimination process, and a pharmacokinetic model incorporating two compartments with capacity-limited elimination from the central compartment was developed to describe the disposition of TG. The TG clearances (derived from model parameters) at the 10- and 20-mg/m2/h dose rates were 987 and 608 ml/min/m2, respectively. Dose-limiting myelosuppression (absolute granulocyte count < 500/mm3 and platelet count < 25,000/mm3) was observed in two of three patients treated with a dose rate of 20 mg/m2/h administered for 36 h. Administration of CIV of TG at 20 mg/m2/h for 24 h was well tolerated in nine patients. Nonhematological toxicities included nonneutropenic infections and mild, reversible changes in hepatic function tests. The recommended dose rate and duration for CIV of TG is 20 mg/m2/h for 24 h.