PURPOSE: To find the more effective treatment methods that improving thesurvival of patients with glioblastoma multiforme(GBM), we analyze the prognostic factors and the outcome of therapy in patients with GBM. METHODS AND MATERIALS: One hundred twenty-one patients with a diagnosis of GBM treated at Severance Hospital between 1973 and 1993 were analyzed for survival with respect to patients characteristics, that is, duration of symptom, age, and Karnofsky performance status, as well as treatment related variables such as extent of surgery and radiotherapy. RESULTS: The median survival time(MST) and 2-year overall survival rate (OSR) of the patients with GBM were 13 months and 20.8%, respectively. Duration of symptom, age, Karnofsky performance status(KPS), radiotherapy, and extent of surgical resection were associated with improved survival in a univariate analysis. Patients whose duration of symptom was longer than 3 months, had the 2-year OSR of 47.2%(p=0.0082), who were younger than age 50, 32.9%(p=0.0003). In patients with a KPS of 80 or higher, the 2 year OSR was 36.9%(p=0.0422). Patients undergoing radiotherapy had the 2-year OSR of 22.9%(p=0.030), and surgical resection of 23.3%(p<0.000). A cox regression model confirmed a significant correlation of duration of symptom, age, radiotherapy,and extent of surgical resection with survival, excluding KPS(p=0.8823). The 2-year OSR were 22.3% and 19.4%, combined with chemotherapy or without, respectively(p=0.06028). The duration of symptom of 3 months or shorter. 50 years of age or older, and undergoing stereotactic biopsy only were considered as risk factors. then patients without any risk factors had the MST of 29 months and 2-year OSR of 53.9% compared to 4 months and 0% for patients who had all 3 risk factors. Most of all treatment failure occurred in the primary tumor site(86.4%) CONCLUSION: The duration of symptom, age, radiotherapy, and extent of surgical resection were a prognostically significant independent variables. To get a better survival, it seems to be reasonable that the study design which improves the local control rates is warranted.