Resection of isolated pulmonary metastases may yield improved survival in select patients. Between 1981 and 1991, 44 women (median age, 55 years) with a history of breast cancer underwent 47 thoracotomies with no operative deaths and only three minor postoperative complications (3/47, 6.4%). Confirmation of the metastatic origin of the lung lesion was made by direct histological comparison with the primary. Three patients had benign nodules and were excluded, and 4 patients had less than complete resection at thoracotomy. The median survival after thoracotomy of the remaining 37 patients with completely resected metastases was 47 +/- 5.5 months, and their actuarial 5-year survival was 49.5%. Patients with a disease-free interval of longer than 12 months had a longer survival (median survival, 82 +/- 6 months; 5-year survival, 57%) than patients with a disease-free interval of 12 months or less (median survival, 15 +/- 3.6 months; 5-year survival, 0%) (p = 0.004). Patients with estrogen receptor-positive status (n = 14) tended to have longer survival after resection than patients with estrogen receptor-negative status (n = 15) (median survival, 81 +/- 9 months versus 23 +/- 6 months, respectively; p = 0.098). Other clinical variables analyzed did not predict survival after thoracotomy. We conclude that resection of pulmonary metastases in patients with breast cancer can be done safely and may result in long-term survival for a substantial number of patients. Patients with a disease-free interval of longer than 12 months have an excellent prognosis after complete resection.