SUMMARY This retrospective analysis involved 453 breast cancer patients treated by radiotherapy alone at the Princess Margaret Hospital and at the Institut Gustave-Roussy. These patients either had operable tumors, but were unfit for general anesthesia, or had inoperable tumors due to local contraindications to surgery. Results were analyzed according to lymph node response, lymph node recurrence rate, distant metastasis rate, survival rate, tumor size, clinical N category, age, tumor dose, axillary dose, and internal mammary chain irradiation. Multivariate analysis permitted the definition of a risk score for axillary lymph node recurrence (RS,,) according to two independent factors: tumor size and clinical N category. It was shown that the RSLN was a good single composite prognostic factor for lymph node control. Increase in axillary dose gave a similar effect on the axillary lymph node recurrence relative risk for all the RSLN groups. According to the slope of the axillary dose-effect curve, it was deduced that a dose increase of 15 Gy can decrease the relative risk of lymph node recurrence 2-fold. A similar slope was previously described for tumor dose and local control. A risk score for distant metastases ( R s d was calculated according to three independent prognostic factors: tumor size, clinical N category, and age. The RSM was not influenced by the tumor or the axillary dose but by the treatment of the internal mammary chain. Similar results were obtained for the risk score for death (RS,). The present data emphasize the role of the lymph node radiation dose in the lymph node control of locally advanced breast cancer.