Background: The Graded Prognostic Assessment (GPA) is a commonly used prognostic index in patients with brain metastases (BM). GPA for melanoma consists of Karnofsky Performance Scale (KPS) and the number of BM present. The purpose of this study was to evaluate the utility of GPA index in a contemporary cohort of patients (pts) with MBM at a single institution to predict Overall Survival (OS). Methods: With IRB approval, the Cleveland Clinic Brain Tumor and Neuro-Oncology Center’s database was used to identify pts with MBM treated between 2000-2012. The primary endpoint was OS from diagnosis of MBM. Cox proportional hazards models were used for data analysis. Stepwise variable selection was used to identify independent prognostic factors. Results: 90 MBM (51 females) median age 57 years (range 24-87) were included for analysis. The median number of BM was 2 (range, 1-11). KPS was 90-100(52%), 70-80 (43%) and <70 (6%). Extracranial metastases was present in 75 patients (83%). Initial treatment included Stereotactic Radiosurgery (SRS) (49%), Whole Brain Radiotherapy (WBRT) (8%), WBRT + SRS (22%), WBRT + Surgery (S)(14%) and SRS + (S) (7%). Median OS was 7.8 months (95% C.I. 6.8-10.1). GPA was prognostic for OS ( p=0.01), however this was because pts with scores of 4 had worse outcomes (median 5.1months) than the other 3 groups, which had similar OS (median 9.0-12.8 months). In addition, number of BM was not associated with OS (p=0.19). In contrast, KPS (p=0.02), Liver (p=0.04) and hemorrhagic metastasis (p=0.02) were independently prognostic for OS. These factors can be used to derive a new prognostic index with 3 groups: Unfavorable, Intermediate and Favorable (see Table). Conclusions: GPA was prognostic for OS in pts with MBM, however separation of the groups was not clear. A new prognostic index consisting of KPS, liver and hemorrhagic metastases is proposed for MBM.