The recent shift toward the diagnosis of smaller breast cancers has led to the reevaluation of their treatment. Because of the low incidence of nodal involvement, recent studies have recommended selective axillary lymph node dissection (AxLND) for early breast cancer. The incidence of nodal involvement is a critical factor in defining the role of AxLND. Large series based on cancer registry data report the incidence of nodal positivity in stage T1a cancer to be 16 to 23 per cent. In contrast, data that include only pathologically reviewed cases report the incidence to be 0 to 6 per cent. We reviewed the medical records of 148 stage T1a breast cancer patients from 1987 through 1994 in two community hospitals as identified by the local tumor registry. After chart review, 115 cases with AxLND underwent pathologic review; 82 were confirmed as stage T1a. Only 3 of 82 (4%) patients were node positive. All three node-positive tumors were of infiltrating ductal histology. No tumor characteristic was predictive of nodal metastasis. Data from the tumor registry and from pathology reports overstated the incidence of nodal involvement (5 and 9%, respectively). In light of the limited clinical benefit and associated cost and morbidity of AxLND, selected informed patients may be spared AxLND.