We used monoclonal antibodies to identify occult micrometastases in the bone marrow of 49 patients with operable (stage I and II) breast carcinoma. Follow-up (mean, 29 months; median, 30 months) revealed that 12 patients recurred. The presence of bone marrow micrometastases (BMM) was significantly associated with early recurrence (P less than .04). The estimated 2-year recurrence rate for patients with no BMM detected (BMM-) was 3%; in patients with BMM, the 2-year recurrence rate was 33%. When BMM and axillary lymph node (LN) status were combined, groups of patients at low risk (LN-, BMM-; 2-year recurrence rate, 0%) and high risk (LN+, BMM+; 2-year recurrence rate, 42%) for early recurrence were identified. Bone marrow tumor burden was related to early recurrence. Among patients with BMM, those who did not recur had on average fewer extrinsic cells in their marrow than those who recurred (15 v 43 cells, respectively). Multivariate analysis comparing BMM, LN+ versus LN-, and tumor size (less than or equal to 2 cm v greater than 2 cm) revealed no factor independently associated with early recurrence. Peripheral tumor burden of BMM (0 or less than 10 extrinsic cells v greater than or equal to 10 extrinsic cells) was the only independent predictor of early recurrence (P less than .003). In conjunction with conventional prognostic factors, particularly axillary LN status, evaluation for BMM might be used to stratify patients for adjuvant treatment programs. Because this pilot study involved few patients with short-term follow-up, the results should be interpreted with caution. The examination of bone marrow for micrometastases remains an experimental procedure; the clinical usefulness of the test will be established through larger studies with long-term follow-up.