Background: Treatment of elderly glioblastoma patients (EGP) is a challenge in neuro-oncology. The comprehensive geriatric assessment (CGA) is currently used to assess geriatric oncological patients with other types of tumors. We performed a large retrospective study to analyze its predictive role in EGP. Methods: Patients aged ≥65 years with histologically confirmed diagnosis of glioblastoma were enrolled. CGA included the following tests: the Cumulative Illness Rating Scale-Comorbidity and Severity Index, Activities of Daily Living, Instrumental Activities of Daily Living, the Mini Mental State Examination, and the Geriatric Depression Scale. Based on CGA results, each patient was categorized as fit, vulnerable, or frail. Results: We enrolled 113 patients. According to the CGA scores, 35% of patients were categorized as "fit", 30% as "vulnerable", and 35% as "frail" patients. Median overall survival was 16.5, 12.1, and 10.3 months in fit, vulnerable, and frail patients (p = 0.1), respectively. On multivariate analysis, the CGA score resulted an independent predictor of survival; indeed, vulnerable and frail patients had a hazard ratio of 1.5 and 2.2, respectively, compared to fit patients (p = 0.04). No association between CGA and progression-free survival (PFS) was demonstrated. Conclusions: The CGA score proved to be a significant predictor of mortality in EGP, and it could be a useful treatment decision tool.