OBJECTIVE The evaluation of radiotherapy and surgery as exclusive local treatment in comparably selected subgroups of patients with Ewing's sarcoma on the basis of the CESS 86-data. METHODS In the German multicenter Ewing's sarcoma study CESS 86, treatment consisted of four 9-week-courses of VACA- or VAIA-chemotherapy plus local therapy. VACA (vincristine, actinomycin D, cyclophosphamide, adriamycin) was given in low-risk extremity tumors with a tumor volume below 100 cm3. High-risk patients with central lesions or a tumor volume > 100 cm3 received VAIA (ifosfamide instead of cyclophosphamide). Local therapy started after one complete chemotherapy course in week 10. Based on an individual decision in each patient, local therapy was either radical surgery or resection plus postoperative irradiation with 45 Gy or definitive radiotherapy with 60 Gy. Because of poor results with radiotherapy in a preceding study, it was intended to restrict irradiation to patients with small lesions. RESULTS Hundred and seventy-seven protocol patients were recruited from January 1986 through June 1991 and 176 received local therapy: 39 underwent radical surgery, 44 received definitive radiotherapy and 93 were treated with resection and postoperative irradiation. The median tumor volume was higher in patients with radiotherapy as compared to combined local treatment or radical surgery, 156 cm3 versus 140 cm3 versus 102 cm3. The overall 5-year survival after radiotherapy and surgery was nearly identical, 63% versus 67% for the whole group 75% versus 65% in tumors < 100 cm3 volume and 65% versus 67% in tumors with 100 cm3 to 600 cm3 volume, respectively. CONCLUSIONS With regard to tumor volume, the most important single prognostic factor in Ewing's sarcoma, irradiated patients were poorer selected than surgically treated patients despite the fact that a selection of good-risk patients for radiotherapy was intended. The nearly identical survival figures after surgery and radiotherapy suggest that radiotherapy is as effective as surgery if selection of patients is comparable.