OBJECTIVE: In elderly patients with severe traumatic brain injury, the use of aggressive versus conservative management remains controversial. The aim of this study was to assess the outcome of surgical alternatives for treatment of severe traumatic intracranial hematoma in patients >= 65 years old and identify factors that may contribute to the outcome. METHODS: This retrospective study included 112 patients with traumatic intracranial hematoma. Logistic regression analysis was used to identify independent predictors of unfavorable outcome. The Glasgow Outcome Scale was used to evaluate the outcome. RESULTS: Surgery was performed in 70 (62.5%) patients >= 65 years old. A favorable outcome occurred in 33 (47.1%) surgical patients, and 23 (32.9%) surgical patients died. Patients who underwent surgery were less likely to have an unfavorable outcome (52.9% vs. 95.2%) and less likely to die (32.9% vs. 88.1%) at 6 months postinjury compared with patients treated conservatively. Multivariate logistic regression analysis revealed that lower Glasgow Coma Scale score (<= 5) was a significant factor associated with unfavorable outcome (odds ratio [OR] = 18.7, 95% confidence interval [CI] = 5.2-95.5, P < 0.001) and mortality (OR = 10.7, 95% CI = 4.4-28.5, P < 0.001). However, neurosurgical intervention was a significant factor to negatively predict unfavorable outcome (OR = 0.03, 95% CI = 0.01-0.1, P < 0.001) and mortality (OR = 0.04, 95% CI = 0.01-0.1, P < 0.001). CONCLUSIONS: This study showed that Glasgow Coma Scale score (<= 5) was a major determinant of outcome in elderly patients with severe traumatic brain injury. However, surgical treatment reduced mortality and improved outcome in the elderly patients in this study.