We retrospectively assessed the predictive factors for brain metastasis in high-risk breast cancer patients receiving radiotherapy after mastectomy. Between January 1994 and 2002, the medical charts of nonmetastatic breast cancer patients receiving post-mastectomy radiotherapy were evaluated. The clinical and pathologic features of patients who developed brain metastasis as the first site metastatic disease were compared with nonmetastatic patients treated at the same time period. All eligible patients in this study were treated with postoperative radiotherapy +/- chemotherapy. Age, stage, percent positive lymph node involvement, number of lymph node metastasis, primary tumor size, grade, surgical margin status, estrogen receptor status, and perinodal fat tissue invasion were analyzed as predictive factors for brain metastasis. Statistical analyses were performed by using Log-rank test and Cox's regression analysis. Median follow-up-time was 61 months. In this period, 32 out of 957 patients (3.3%) developed brain metastasis. In univariate analysis percent positive axillary lymph node involvement (p < 0.001), primary tumor size (p < 0.001), number of lymph node metastasis (p = 0.01), and American Joint Committee on Cancer 2002 stage (p < 0.001) were found to be predictive for brain metastasis. Multivariate analysis revealed that only the primary tumor size and percent positive lymph node involvement were significant predictive factors for the development of brain metastasis. The primary tumor size and percent positive lymph node involvement increases the risk of brain metastasis in patients with nonmetastatic breast cancer receiving postoperative radiotherapy and chemotherapy.