Purpose: To ascertain predictors of distant brain failure (DBF) in patients treated initially with stereotactic radiosurgery alone for newly diagnosed brain metastases. We hypothesize that these factors may be used to group patients according to risk of DBF. Methods and Materials: We retrospectively analyzed 100 patients with newly diagnosed brain metastases treated from 2003 to 2005 at our Gamma Knife radiosurgery facility. The primary endpoint was DBF. Potential predictors included number of metastases, tumor volume, histologic characteristics, extracranial disease, and use of temozolomide. Results: One-year actuarial risk of DBF was 61% for all patients. Significant predictors of DBF included more than three metastases (hazard ratio, 3.30; p = 0.004), stable or poorly controlled extracranial disease (hazard ratio, 2.16; p = 0.04), and melanoma histologic characteristics (hazard ratio, 2.14; p = 0.02). These were confirmed in multivariate analysis. Those with three or fewer metastases, no extracranial disease, and nonmelanoma histologic characteristics (N = 18) had a median time to DBF of 89 weeks vs. 33 weeks for all others. One-year actuarial freedom from DBF for this group was 83% vs. 26% for all others. Conclusions: Independent significant predictors of DBF in our series included number of metastases (more than three), present or uncontrolled extracranial disease, and melanoma histologic characteristics. These factors were combined to identify a lower risk subgroup with significantly longer time to DBF. These patients may be candidates for initial localized treatment, reserving whole-brain radiation therapy for salvage. Patients in the higher risk group may be candidates for initial whole-brain radiation therapy or should be considered for clinical trials. © 2008 Elsevier Inc. All rights reserved.