Previous uncontrolled clinical trials have shown the in vivo retrovirus (RV)-mediated transduction of glioblastoma cells with the herpes simplex virus thymidine kinase (HSV-tk) gene and subsequent systemic treatment with ganciclovir to be feasible and well tolerated. However, because of continued tumor progression in most patients, the antitumor effect could not be determined using historical controls. Here, we describe a phase III, multicenter, randomized, open-label, parallel-group, controlled trial of the technique in the treatment of 248 patients with newly diagnosed, previously untreated glioblastoma multiforme (GBM). Patients received, in equal numbers, either standard therapy (surgical resection and radiotherapy) or standard therapy plus adjuvant gene therapy during surgery. Progression-free median survival in the gene therapy group was 180 days compared with 183 days in control subjects. Median survival was 365 versus 354 days, and 12-month survival rates were 50 versus 55% in the gene therapy and control groups, respectively. These differences were not significant. Therefore, the adjuvant treatment improved neither time to tumor progression nor overall survival time, although the feasibility and good biosafety profile of this gene therapy strategy were further supported. The failure of this specific protocol may be due mainly to the presumably poor rate of delivery of the HSV-tk gene to tumor cells. In addition, the current mode of manual injection of vector-producing cells with a nonmigratory fibroblast phenotype limits the distribution of these cells and the released replication-deficient RV vectors to the immediate vicinity of the needle track. Further evaluation of the RV-mediated gene therapy strategy must incorporate refinements such as improved delivery of vectors and transgenes to the tumor cells, noninvasive in vivo assessment of transduction rates, and improved delivery of the prodrug across the blood-brain and blood-tumor barrier to the transduced tumor cells.