Glioblastoma multiforme (GBM) is one of the most aggressive human tumors with poor prognosis. Current standard treatment includes chemotherapy using DNA alkylating agent temozolomide (TMZ) concomitant with surgical resection and/or irradiation. However, GBM patients exhibit various levels of the elevated expression of DNA repair enzyme, due to MGMT causing resistance to TMZ. Determination of the MGMT-positive population of primary tumor is important to evaluate the therapeutic efficacy of TMZ. Here we generated TMZ-resistant GBM cells by introducing MGMT into TMZ-sensitive GBM cell line KMG4, and established a model to assess the TMZ-induced bystander effect on TMZ-resistant cells. By mixing TMZ-resistant and -sensitive cells, GBM tumors with MGMT positivity as 50%, 10%, and 1% were generated in vivo. We could not observe any bystander effect of TMZ-induced cell death in tumor with 50% MGMT positivity. Although the bystander effect was observed within 20days in the case of tumor with 1% MGMT positivity, final tumor size at day 28 was the same as control without sensitive cells. This bystander effect was observed in vitro using conditioned medium of TMZ-damaged GBM cells, and PCR array analysis indicated that the conditioned medium stimulated stress and toxicity pathway and upregulated anti-oxidants genes expression such as catalase and SOD2 in TMZ-resistant cells. In addition, the reduction of the activity of anti-stress mechanism by using inhibitor of GSH synthesis potentiated TMZ-induced bystander effect. These results suggest that GSH inhibitor might be one of the candidates for combination therapy with TMZ for TMZ-resistant GBM patients.