In preparation for a phase I clinical trial using a combined cytotoxic/immunotherapeutic strategy with adenoviruses (Ad) expressing Flt3L (Ad-Flt3L) and thymidine kinase (Ad-TK) to treat glioblastoma (GBM), we tested the hypothesis that Ad-TK+GCV would be the optimal tumor-killing agent in relation to efficacy and safety when compared with other proapoptotic approaches. The efficacy and neurotoxicity of Ad-TK+GCV was compared with Ads encoding the proapoptotic cytokines tumor necrosis factor-alpha, tumor necrosis factor-related apoptosis-inducing factor (TRAIL), and Fas ligand (FasL), alone or in combination with Ad-Flt3L. In rats bearing small GBMs (day 4), only Ad-TK+GCV or Ad-FasL improved survival. In rats bearing large GBMs (day 9), the combination of Ad-Flt3L with Ad-FasL did not improve survival over FasL alone, whereas Ad-Flt3L combined with Ad-TK+GCV led to 70% long-term survival. Expression of FasL and TRAIL caused severe neuropathology, which was not encountered when we used Ad-TK+/-Ad-Flt3L. In vitro, all treatments elicited release of high mobility group box 1 protein (HMGB1) from dying tumor cells. In vivo, the highest levels of circulating HMGB1 were observed after treatment with Ad-TK+GCV+Ad-Flt3L; HMGB1 was necessary for the therapeutic efficacy of AdTK+GCV+Ad-Flt3L because its blockade with glycyrrhizin completely blocked tumor regression. We also showed the killing efficacy of Ad-TK+GCV in human GBM cell lines and GBM primary cultures, which also elicited release of HMGB1. Our results indicate that Ad-TK+GCV+Ad-Flt3L exhibit the highest efficacy and safety profile among the several proapoptotic approaches tested. The results reported further support the implementation of this combined approach in a phase I clinical trial for GBM.