Between 1975 and 1982, 80 patients with osteogenic sarcoma were entered into prospective trials in the Surgery Branch of the National Cancer Institute. In 43 of these patients, pulmonary metastases developed as the initial site of recurrence, and 39 underwent one or more thoracotomies for resection of the disease. The actuarial five-year survival for the group of 43 patients with pulmonary metastases was 40%. Various prognostic factors were analyzed for their influence on survival after thoracotomy. Age, sex, location of primary tumor, tumor doubling time, and involvement of one or both lungs (bilaterality) were not significant in predicting survival. Prognostic factors that influenced survival, calculated by regression analysis, included the number of nodules on preoperative lung tomograms (negative correlation, p = 0.0004), disease-free interval (positive correlation, p = 0.0136), resectability (positive correlation, p = 0.002), and the number of metastases resected at thoracotomy (negative correlation, p = 0.0032). The presence of 3 nodules or less on preoperative full-lung linear tomography was found to be the single most useful preoperative prognostic factor. The application of these prognostic factors preoperatively may identify patients who will benefit optimally from thoracotomy.