Although previous studies have indicated that the predictors of local recurrence following conservative surgery (CS) and radiotherapy (RT) are not the same as those following mastectomy, it remains unclear whether the predictors of distant relapse differ by local treatment modality. Clinical and pathologic features predictive of distant relapse for patients treated with mastectomy have been well established and include lymph node involvement, histologic grade, and peritumoral lymphatic vessel invasion (LVI). To study the influence of these and other factors on the rate of distant relapse in patients treated with CS and RT, we have identified a group of 438 patients treated between 1968 and 1981 who met the following criteria: primary tumor size less than or equal to 5 cm, excision of the primary tumor, infiltrating ductal carcinoma as the most aggressive histologic subtype, histology evaluable for the presence of an extensive intraductal component, and a dose to the primary site greater than or equal to 60 Gy. Estrogen receptor status was available in 58% of cases, 76% had an axillary dissection, and 23% were treated with adjuvant chemotherapy. With a median follow-up of 89 months, 107 patients (24%) developed a distant relapse. The 5-year actuarial freedom from distant relapse (FDR) was 80%. Stepdown Cox proportional hazards regression analysis identified several factors associated with a significantly (p less than 0.01) increased risk for distant relapse: positive lymph nodes, histologic grade, necrosis, and lymphatic vessel invasion. The magnitude of each effect was then examined with a lifetable calculation. Five-year freedom from distant relapse was 86% for the node-negative subgroup, 78% for patients with one to three positive nodes, and 45% for patients with four or more positive nodes. For histologic grades I, II, and III, 5-year freedom from distant relapse was 96%, 97%, and 75%, respectively. For necrosis scored as absent, scant, moderate, or marked, 5-year freedom from distant relapse was 90%, 78%, 77%, and 66%, respectively. For lymphatic vessel invasion scored as absent or present, 5-year freedom from distant relapse was 85% and 63%, respectively. We conclude that the clinico-pathologic predictors for distant relapse following conservative surgery and radiotherapy appear to be the same as those following mastectomy. This observation is consistent with the notion that distant relapse is caused by the presence of micrometastases at the time of initial patient sentation and is not greatly influenced by selection of local treatment.