Although more and more patients with Ewing's sarcoma of bone (ESB) are being treated by surgery, the relative role of surgery and radiotherapy in the local treatment of this tumor has yet to be determined. Because the outcome of ESB may differ according to the anatomical site of the tumor, results reported in the literature, which generally refer to series with tumors located in all sites, may be selection biased. Therefore, we have retrospectively evaluated patients with ESB exclusively in the extremity and locally treated by surgery or radiotherapy. Two hundred and sixty-eight patients treated at Rizzoli 1979-1996 for non-metastatic ES of the extremities were assessed. Chemotherapy was administered according to four sequentially activated protocols. One hundred and thirty-six patients were treated by surgery, 70 by surgery and radiotherapy, and 60 patients by radiotherapy. Two patients underwent only chemotherapy. The follow-up range was 5-23 years (mean 13 years). One hundred and fifty-two patients remained continuously free of disease, 108 relapsed, 2 died of chemotherapy toxicity and 6 developed a second malignancy. The 5-year event-free survival (EFS) and overall survival (OS) were respectively 62 and 69%. Although patients of all groups were matched for possible risk factors, the rates of 5-year EFS and local control were significantly lower in patients treated with radiotherapy compared to patients treated by surgery or surgery and radiotherapy (48% vs 66%, p=0.002; 80% vs 94%, p=0.0001). Furthermore, in group 3 there were 6 secondary malignancies. Our results indicate that surgery should always be considered in the local treatment of ES of the extremities. Postoperative radiation therapy must be added in case of inadequate surgical margins.