We prospectively analyzed the safety and outcome of frameless image-guided robotic stereotactic radiosurgery (SRS) for treatment of brain metastases in patients that would have otherwise been treated with frame-based techniques. During a three-year period, 333 patients with 783 brain metastases of various histologies underwent 391 outpatient SRS procedures. Fifty-five percent of patients had multiple brain metastases. The median (mean) tumor volume was 1.0 cc (2.7 cc). The mean prescribed tumor dose was 18.5 Gy (+/-1.3 Gy). Local/distant tumor recurrences were treated by additional SRS for patients with stable systemic disease. Survival and freedom from local tumor recurrence was analyzed with the Kaplan-Meier method. Prognostic factors were obtained from the Cox proportional hazards model. System accuracy tests (end-to-end tests) were performed with a standard head phantom. Overall median survival was 12.2 months after SRS. The actuarial one-year local control rate was 95.2% (95% CI: 92.0-97.2); the distant brain tumor control rate was 67% (95% CI: 61.0-71.2). Most patients died from systemically progressing cancer (69%). A Karnofsky performance score (KPS) > 70 was related to prolonged survival in the univariate and multivariate analysis. Recursive partition analysis (RPA) classes I and II were related to prolonged survival in the univariate analysis. Twenty-one patients (6.3%) developed treatment-related neurotoxic effects; no patient died because of complications of SRS. Forty-five end-to-end tests documented a mean targeting accuracy of 0.48 +/- 0.22 mm. Single-session, frameless robotic SRS is feasible, accurate, and safe in selected patients with brain metastases of various primary tumors. There seems to be no difference in patient selection, adverse effects, treatment outcomes, or system accuracy compared with frame-based SRS.