Forty-seven adult patients with glioblastoma multiforme (GBM) were treated in a phase I/II study combining continuous intravenous (IV) infusions of iododeoxyuridine (IdUrd) and hyperfractionated radiation therapy. IdUrd was administered as a continuous infusion (24 h/d) for two separate 14-day infusion periods. The dose of IdUrd was escalated from 500 to 1,200 mg/m2/d. The initial wide-field tumor volume was treated to 45 Gy at 1.5 Gy fractions twice daily over 3 weeks. Following a planned 2-week break, a reduced-field boost of 25 Gy was delivered using 1.25 Gy fractions twice daily over 2 weeks (total dose, 70 Gy over 9 weeks). The IdUrd infusion preceded both the wide-field and reduced-field irradiation by 1 week. All treatment was performed on an outpatient basis. Dose-limiting systemic toxicity to the bone marrow (primarily thrombocytopenia) and gastrointestinal (GI) tract (both stomatitis and diarrhea) established the maximum tolerable dose (MTD) at 1,000 mg/m2/d for a 14-day infusion. Significant local toxicity (within the radiation field) was not seen. The kinetics of IdUrd were linear with dose escalation and reached steady-state plasma concentrations of 1.3 to 3.4 mumol/L. The total body clearance of IdUrd was .82 L/min/m2. The primary metabolite, 5-iodouracil (IUra) approached steady state by day 6 of the infusion when plasma levels were 60 times higher than IdUrd. Plasma levels of uracil and thymine, but not thymidine, were elevated throughout the infusion. With a minimum follow-up of 1 year, ten patients remain alive, while 33 patients died of progressive disease, and four patients died of other causes (including one treatment-related death). The median survival for all 47 patient and for the 40 patients receiving the MTD was 45 and 47 weeks, respectively, with 12% and 14% survivals at 24 months. Using a Cox regression analysis, age (less than or equal to 50 years v greater than 50 years) and pretreatment performance status (PS) (Eastern Cooperative Oncology Group [ECOG]-PS 0 to 1 v PS 2 to 3) were independent, statistically significant (P2 less than .05) predictors of survival, with the ECOG status being a better predictor. Patients with a PS 0 to 1 (28 patients) had a median survival of 64 weeks with 21% survival at 24 months, compared with a median survival of 29 weeks and 0% survival at 12 months in the 19 patients with PS 2 to 3. The overall and subgroup survival data are at least comparable to other combined modality treatment approaches in patients with GBM.