Central nerve system (CNS) metastases are a feared complication of breast cancer and are associated with poor prognosis. The purpose of this study is to investigate the clinical characteristics of CNS metastases and to clarify the prognostic factors after CNS metastases in breast cancer at a single institution over a long time period. We retrospectively reviewed the medical records of breast cancer patients diagnosed at Seoul National University Hospital from 1981 to 2009 and identified the patients who experienced CNS metastases. We collected the data, including demographics, clinico-pathologic characteristics, dates of diagnosis of original breast cancer and subsequent metastases, and date of death, and correlated the findings with the clinical outcome. A total of 400 patients were identified, of whom 17 (4.3%) were diagnosed with CNS metastases and primary breast cancer concurrently and 383 (95.7%) experienced CNS metastases subsequent to the diagnosis of primary breast cancer. Further, 318 patients (79.5%) had only brain parenchymal metastases, 30 (7.5%) had only leptomeningeal metastases, and 52 (13%) had both. After the diagnosis of CNS metastasis, 170 patients (42.5%) received systemic chemotherapy (CTx) and 143 (35.8%) received CTx after whole brain radiation therapy (WBRT). The patients with good performance status (PS), initial CNS metastasis as recurrence, absence of extracranial metastases, non-visceral extracranial metastases, longer interval from the date of primary breast cancer to the date of CNS metastasis, and CTx after WBRT and gamma-knife surgery (GKS), had better outcomes in univariate analyses. In multivariate analysis, good PS, systemic CTx after WBRT, GKS, and longer interval to CNS metastasis, were independent prognostic factors for overall survival after CNS metastases. Our results suggest that appropriate palliative systemic therapy after WBRT or GKS, and adequate palliative treatment via combined modalities are helpful for breast cancer patients, even after the detection of CNS metastases.