OBJECTIVE To determine if involved field radiation (IF) is equivalent to standard whole bone radiation (SF) in local tumor control; to establish patterns of failure following treatment; and to determine response, event-free survival (EFS), and overall survival rates from multidisciplinary therapy in Ewing's sarcoma. METHODS Between 1983 and 1988, 184 children with Ewing's sarcoma were enrolled onto Pediatric Oncology Group 8346 (POG 8346). A total of 178 (97%) met eligibility criteria; 6 had pathology other than Ewing's sarcoma. Induction chemotherapy of cyclophosphamide/doxorubicin (adriamycin )(C/A) x 12 weeks was followed by local treatment either surgery or radiation therapy and C/A, dactinomycin, and vincristine for 50 weeks. Resection was advised for patients with small primary tumors if accomplished without functional loss. Forty patients were randomized to receive SF, whole bone radiation to 39.6 Gy plus a 16.2 Gy boost (total 55.8 Gy) or IF to 55.8 Gy, and the remainder were assigned to IF radiation. RESULTS Of 178 eligible patients, 141 (79%) had localized disease and 37 (21%) had metastases at presentation. Their 5-year EFS was 51% (SE 5%) and 23% (SE 7%) respectively. The response rate to induction chemotherapy was 88% (28% complete, 60% partial), but after radiotherapy the response rate increased to 98%. Thirty-seven of the localized patients underwent resection, of whom 16 (43%) required postoperative radiotherapy; the 5-year EFS of these surgical patients was 80% (SE 7%). The remaining 104 localized patients were eligible for randomization or assignment to receive radiotherapy; the 5-year EFS of these patients was 41% (SE 5%), with no significant difference in EFS between those randomized to SF vs. IF. Site of primary tumor correlated with 5-year EFS: distal extremity 65% (SE 8%), central 63% (SE 10%), proximal extremity 46% (SE 8%), and pelvic-sacral 24% (SE 10%) (p=0.004). Initial tumor size did not correlate significantly with EFS. Patterns of failure among the 141 localized patients revealed 23% of patients experienced a local failure, while 40% had a systemic failure. The 5-year local control rate for the surgical patients +/- postoperative radiotherapy was 88% (SE 6%), while for the patients undergoing radiotherapy alone it was 65% (SE 7%). There was no difference in local control between those randomized to SF vs. IF. The 5-year local control rate for the patients with pelvic-sacral tumors was 44% (SE 15%), significantly worse than the local control rates for those with central tumors 82% (SE 8%), distal extremity 80% (SE 8%), or proximal extremity 69% (SE 9%) (p=0.023). However, quality of radiotherapy correlated with outcome. Patients who had appropriate radiotherapy had a 5-year local control of 80% (SE 7%), while those with minor deviations had 5-year local control of 48% (SE 14%), and those with major deviations had a local control of only 16% (SE 15%) (p=0.005). The local failure was within an irradiated volume in 62% of patients, outside the irradiated volume in 24% of cases, while the precise location could not be determined in the remaining 14%. CONCLUSIONS As most failures in Ewing's sarcoma are systemic, improved EFS requires more effective systemic chemotherapy. Adequate IF radiotherapy requires treatment to appropriate volumes as defined by MRI imaging and full radiation doses. Pretreatment review of radiologic images with a musculoskeletal radiologist to determine appropriate tumor volumes, as well as use of conformal radiotherapy techniques are important for improved outcome.