Background: A previously described diagnosis-specific graded prognostic assessment (DS-GPA) for melanoma showed that number of brain metastases (BM) and Karnofksy Performance Score (KPS) were significant prognostic factors. The purpose of this study was to test this DS-GPA in our patient population undergoing stereotactic radiosurgery (SRS) and to determine if other prognostic factors could enhance the prognostic accuracy and utility of this index. Methods: All patients who underwent SRS for melanoma BM from 1999 to 2009 were identified. Demographic, disease and survival data were collected. Continuous variables were stratified according to previously published prognostic models (age <50, 51-59, ≥60; number of BM1, 2-3, >3; KPS <70, 70-80, 90-100). The primary endpoint was overall survival (OS), calculated using Kaplan-Meier method with log rank and multivariate (Cox) analysis. Results: 126 patients underwent SRS for newly diagnosed melanoma BM. Median OS was 8.1 months (range 1.0 to 143.0 months). Increasing number of BM (10.7 versus 7.2 versus 4.9, p=0.030), LDH greater than 200 (11.4 versus 5.4, p=0.032) and decreased KPS (12.4 versus 7.1 versus 5.6, p=0.007) correlated with worse survival, but age, extracranial metastases, and treatment with temozolomide, whole brain radiation or resection did not. On multivariate analysis KPS (HR 2.89, CI 1.34-6.25) and LDH (HR 1.71, CI 1.03-2.83) maintained significance. When analyzed with only number of BM and KPS, the two validated DS-GPA prognostic factors, LDH remained a negative prognostic factor (HR 1.73, CI 1.06- 2.83). Conclusions: If confirmed in a large prospective study, the addition of LDH may improve the prognostic accuracy and utility of a previously described prognostic index based on number of BM and KPS for patients with melanoma metastatic to the brain.