From July 1975 to December 1982, 358 patients were referred to the Surgery Branch of the National Cancer Institute (NCI) with the diagnosis of soft tissue sarcoma of the extremities. Two hundred eleven of these patients presented with resectable, localized high-grade soft-tissue sarcomas and have been included in the present analysis of the management and outcome of patients with high-grade soft tissue sarcomas of the extremity treated at the NCI. One hundred forty-seven of these 211 patients have been included in randomized prospective trials. The remaining 64 patients in this analysis have been followed at the NCI, but were not included in randomized trials because of patient refusal or ineligibility. Tumor size was identified as a highly significant prognostic variable for disease-free and overall survival (P2 = 0.00001 and 0.0081, respectively). Tumor site, histologic type, and microscopic margins of resection were not significant prognostic variables. There was no difference between patients undergoing amputation compared to those undergoing limb-sparing procedures plus postoperative radiotherapy in disease-free or overall survival for all 211 patients in this study (P2 = 0.068 and 0.131, respectively). A significantly greater frequency of local failure among patients treated by local excision was noted compared to patients undergoing amputation (12/128 versus 0/83, P2 = 0.004), but this did not result in decreased overall survival in patients undergoing combined modality limb-sparing procedures. Adjuvant chemotherapy significantly prolonged disease-free survival (P2 = 0.005) for the 124/211 patients treated with adjuvant chemotherapy, although analysis of overall survival did not reveal a significant increase (P2 = 0.10). In a subset of 65 patients included in a prospective randomized trial evaluating the efficacy of adjuvant chemotherapy, a significant improvement in both disease-free (P2 = 0.033) and overall (P2 = 0.055) survival was seen in patients receiving chemotherapy. Sixty-five patients developed recurrent disease (65/211, 31%) and 42 of these patients were rendered disease-free surgically. Survival from the time of first recurrence was significantly prolonged among the 42 patients who were rendered disease-free (median survival, 31 months) compared to those who were not (median survival, 9 months, P2 less than 0.001).