The optimal extent of axillary dissection in patients with breast cancer remains unclear. We report 278 total axillary lymphadenectomies (levels I, II, and III and Rotter's [interpectoral] nodes) that were performed in 264 closely followed up private patients. There have been no axillary recurrences to date (mean follow-up, 50 months). If only level I and II nodes had been removed, the false-negative staging error would have been only 2.6%. However, 29 (31.5%) of 92 pathological node-positive axillae contained apical and/or Rotter's metastases. The incidence of complications was comparable with that reported for partial lymphadenectomy. Arm lymphedema developed in 6% of nonirradiated patients; postoperative radiotherapy and gross nodal disease were significant risk factors for lymphedema. Total axillary lymphadenectomy largely prevents axillary, recurrence, eliminates the small staging error inherent in partial lymphadenectomy, and has acceptable morbidity, provided radiotherapy to the regional nodal areas is avoided.