Risk factors for isolated local-regional (LR) recurrence following mastectomy for breast cancer were analyzed in a review of 627 women entered into Eastern Cooperative Oncology Group (ECOG) adjuvant chemotherapy trials between 1978 and 1982. Premenopausal patients were randomized to cyclophosphamide, methotrexate, and fluorouracil (5-FU) (CMF), cyclophosphamide, methotrexate, 5-FU, and prednisone (CMFP), or cyclophosphamide, methotrexate, 5-FU, prednisone, and tamoxifen (CMFPT). Postmenopausal patients were randomized to observation, CMFP, or CMFPT. Median follow-up time was 4.5 years. At 3 years, 225 patients relapsed and in 70 (31% of failures, 11% of all patients) the initial site was LR without distant metastases. In a multivariate analysis, the risk of an isolated LR recurrence significantly correlated with the number of positive axillary nodes, the primary tumor size, the presence of tumor necrosis, and the number of axillary nodes examined. Factors that significantly discriminated between an isolated LR recurrence and distant metastasis were the number of positive nodes and primary tumor size. Patients with four to seven positive nodes or tumor size greater than or equal to 5 cm had a chance of developing an isolated LR recurrence almost equal to the risk of distant metastases. These findings suggest a potential for improved survival in this subset of patients with the addition of postmastectomy radiation to chemotherapy, and continue to emphasize the presence of a group of patients at high risk for isolated LR recurrence despite adjuvant chemotherapy.