PURPOSE: This study was performed to determine the incidence of BMM and to correlate the presence of these micrometastases with prognosis and other clinicopathologic features. MATERIALS AND METHODS: BMM was evaluated in 220 breast cancer patients between July, 1991 and January, 1997, using mouse monoclonal antibody (AEI/AE3) against cytokeratin in an immunofluorescent assay. RESULTS: Of the 220 patients, 71 (32.3%) were positive for BMM. There were no association between bone marrow positivity and nodal status, TNM stage, known histopathologic parameters, and hormonal receptor, Median follow-up for 220 patients was 41.6 month. The relapse rate was 16.8% (37/220). Twenty-four (33.8%) of 37 patients were positive for BMM and 13 (8.7%) were negative (p<0.05). Bone metastasis occurred in 16 cases, and was more common in BMM positive patients (14 of 24, 54.2%, versus 2 of 13, 15.4%, p<0.05). Twenty-six patients were died of relapsed breast cancer. In overall survival, patients who was negative for BMM showed higher survival rate (p=0.05). CONCLUSIONS: BMM was a good predictor for distant metastasis, especially bone matastasis, and for poor prognosis. But no association was found between bone marrow positivity and tumor size, nodal status, stage, histologic parameter and hormonal receptor status.