PURPOSE: To investigate the degree and effect of cyclooxygenase (COX)-2 expression on the survival of patients with glioblastoma multiforme (GM). MATERIALS AND METHODS: Between 1997 and 2006, thirty consecutive GM patients treated with surgery and postoperative radiotherapy (dose range: 44~65.1 Gy, median dose: 61.2 Gy) were included in the study. Three patients were excluded that discontinued radiotherapy before receiving a dose of 40 Gy due to mental deterioration. The expression of the COX-2 protein in surgical specimens was examined by immunohistochemical analysis. Survival analysis and verification were performed with respect to sex, age, performance status, resection extent, radiotherapy dose, and degree of COX-2 expression using the Kaplan-Meier method and the log rank test. RESULTS: The median length of follow-up was 13.3 months (range: 6~83 months). Staining for COX-2 was positive in all patient samples. Staining for COX-2 that was positive for over 75% of the tumor cells was found in 24 patients. Staining for COX-2 that was positive in less than 25% of tumor cells was found in 3 patients (10.0%), staining for COX-2 that was positive in 25 to 50% of tumor cells was found in 1 patient (3.3%), staining for COX-2 that was positive in 50 to 75% of tumor cells was found in 2 patients (6.7%) and staining for COX-2 that was positive in 75 to 100% of tumor cells was found in 24 patients (80.0%). The median survival and two-year survival rate were 13.5 months and 17.5%, respectively. The survival rate was influenced significantly by the degree of resection (tumor removal by 50% or more) and radiotherapy dose (59 Gy or greater) (p<0.05). The median survival of patients with staining for COX-2 that was positive in less than 75% of tumor cells and in at least 75% of tumor cells was 15.5 and 13.0 months, respectively (p>0.05), and the two-year survival for these groups was 33.3 and 13.3%, respectively (p>0.05). CONCLUSION: The absence of a statistical correlation between the degree of COX-2 expression and survival in GM patients, despite the high rate of COX-2 positive tumor cells in the GM patient samples, requires further studies with a larger series to ascertain the prognostic value of the degree of COX-2 expression in GM patients.