We have analyzed a series of 225 patients with intracranial metastases (343 lesions), treated in our department by Gamma Knife radiosurgery over a 30-month period. We have used a modified Pittsburgh protocol and performed 242 procedures on 164 single/78 multiple lesions. Primary tumors were mostly carcinomas of the lung (52%) and breast (11.6%). Neuroradiological localization of the target was usually performed by stereotactic computed tomography. Magnetic resonance imaging was only used in special circumstances. Routine dose planning was assisted by three-dimensional reconstruction programs. Mean tumor volume was larger than expected (5.7 ml). Mean prescription dose and average dose were 21.1 and 29.9 Gy, respectively. Middle- and long-term results were evaluated in a subset of 152 patients (236 lesions) with adequate (> 4 months) follow-up. Mean follow-up was 53.1 weeks with 61/152 patients still living. There was a predominance of retrospectively classified 'not fully eligible cases' among the survivors, mainly because of uncontrolled primary tumor. The 1-year local tumor control rate was 88.2%. Treatment-related radiological (3.9%) and clinical (1.6%) sequelae were minimal. Overall mean survival in these patients (40 weeks) turned out to be higher than that commonly reported after conventional surgical-radiation treatments. It was encouraging that the mean survival of 'fully' eligible patients was 51 weeks. Karnofsky performance status and neurological (Order Grading) performance scores were consistently high for most of the follow-up period. Functional Independence and the Palliative Index were not far from the value of mean survival. The main cause of death remains uncontrolled systemic disease (64.8%). On the other hand, the relative incidence of intracranial tumor progression was considerably decreased. This indicated that these patients should perhaps be treated more aggressively and underlines the need for randomized trials to determine the optimal treatment.