The optimal extent of breast resection before irradiation for treatment of early breast cancer has not been defined. Increasing the size of the resection may decrease the risk of local recurrence but will also have an adverse impact on the cosmetic outcome. The 5-year likelihood of a recurrence of the tumor was analyzed in relation to the volume of resected breast tissue in 507 patients with infiltrating ductal carcinoma treated with conservative surgery and radiation therapy between 1968 and 1982. Patients were stratified by clinical T-stage and for each T-stage patients were divided into three groups of equal numbers based on the volume of excised tissue. All patients had at least a gross excision of the tumor and the extent of breast resection was determined at the discretion of the surgeon without knowledge of the histologic features of the tumor. The median follow-up time was 100 months. The 5-year actuarial recurrence rates were analyzed in relation to clinical T-stage (T1 or T2) and the presence or absence of an extensive intraductal component (EIC+ or EIC-). For patients with EIC+ tumors, the largest resections were associated with a substantially lower risk of recurrence in the breast than the smallest resections. This effect was seen both for T1 tumors (10% versus 29%, p = 0.07) and for T2 tumors (9% versus 36%, p = 0.04). For patients with EIC-tumors, recurrence rates were significantly lower than for EIC+ tumors and were not influenced by the volume of resection to the same degree as EIC+ tumors. In the absence of an EIC, recurrence rates for the largest and smallest resections were 0% and 9% (p = 0.02) for T1 tumors and 3% and 6% (p = NS) for T2 tumors. It is concluded that a limited breast resection is acceptable for an EIC- tumor but that a more extensive resection is required for an EIC+ tumor. These results stress the importance of assessing the presence or absence of an EIC in determining the optimal extent of breast resection required before radiation therapy.