Patient characteristics of 272 patients entered in a clinical trial conducted by the Pediatric Intergroup Ewing's Sarcoma Committee between June 1972 and November 1978 were examined for their relationship to prognosis. Prognosis was defined as disease-free survival time (time to local recurrence and/or metastatic disease) and overall survival time; all times were measured from the start of treatment. In a multivariate regression model, primary site of disease was the major variable that influenced prognosis, and patients with pelvic sites had the least favorable prognoses, followed by those with proximal and rib sites. The most favorable sites were distal and other. The median disease-free and survival times in weeks by primary site were, respectively: pelvis (69, 112), proximal (102, 141), rib (105, 109+), distal (226+, 240), and other (96+, 199+). Females had better prognoses than males; the median survival times were 197 and 147 weeks, respectively. An abnormal liver function as indicated by an abnormal serum glutamic-oxaloacetic transaminase value (greater than 45 IU) was a bad prognostic sign, although only 8 patients had this finding; their median survival time was 94 weeks. Patients who had resections had a slight advantage in survival compared with those having biopsies, though the difference favoring resection patients was not consistent for both sexes in any primary site. Individual characteristics of the patients that were of prognostic significance were: blood lymphocyte counts (high counts favorable), polymorphonuclear leukocyte counts (high counts unfavorable), and time from symptoms to diagnosis (times less than 1 mol favorable). Patients who received treatment 2 had significantly poorer prognoses than those given treatments 1 or 3. The median disease-free and survival times by treatment were (in wk): 1 (134, 198+), 2 (81, 120), and 3 (123, 182).