Background: 5-year survival of RR-ES is about 15%. rEECur, the first randomized controlled trial in this setting, is defining standard care, balancing efficacy and toxicity. Methods: Patients aged 4-50 with RR-ES were randomly assigned to topotecan and cyclophosphamide (TC), irinotecan and temolozomide (IT), gemcitabine and docetaxel (GD), or high-dose ifosfamide (IFOS). Primary outcome was event-free survival (EFS) for the phase III comparison. Secondary outcomes included overall survival (OS), toxicity, and quality of life (QoL). A probability-based Bayesian approach was used with multiple pairwise comparisons. At the first and second interim assessments, patients allocated to GD and IT, respectively, had worse objective response (OR) and EFS than the other arms, halting recruitment to both. The final intent-to-treat assessment of the original four arms was a phase III evaluation of TC and IFOS. Results: 451 patients recruited between 18/12/14 and 31/08/21, were randomly assigned to TC (163 patients), IT (127 patients), GD (72 patients), and IFOS (83 patients). Median age was 19 years (range 4-49). Patients had: refractory disease (18%), first recurrence (66%), > first recurrence (17%). Initial disease site was bone in 70%. Sites of progression were: primary site only (15%), pleuropulmonary metastases only (34%), and other metastatic (51%). Baseline renal function was similar in both. Median follow-up (reverse Kaplan-Meier method) was 40 months. For the phase III comparison between TC and IFOS (both, 73 patients), median EFS was 3.7 months (95% CI, 2.1-6.2) for TC and 5.7 months (95% CI, 3.8-7.0) for IFOS. Median OS was 10.4 months (95% CI, 7.5-15.5) for TC and 16.8 months (95% CI, 11.1-25.8) for IFOS. Given the observed data, the posterior probability that EFS and OS were better after IFOS than after TC (ie Pr [true hazard ratio < 1 | data]) was 95% for both. A greater survival difference was observed for patients aged under 14 than those aged ≥ 14 for EFS and OS. Subgroup analyses favored IFOS for all minimization factors. The main grade 3/4 adverse events (% patients with an event) for TC (left-hand values) compared with IFOS were: febrile neutropenia (26% vs. 25%), infections (8% vs. 14%), vomiting (1% vs. 1%), nausea (0% vs. 3%), diarrhea (1% vs. 1%), encephalopathy (0% vs. 7%), and renal toxicity (0% vs. 8%). Descriptive statistics of quality of life scores appeared to favor the IFOS arm over the TC arm in children but not in adults. Conclusions: The first randomized trial in RR-ES has shown that high-dose ifosfamide is more effective in prolonging survival than TC, having previously beaten GD and IT, and should be considered as a control arm in future randomized phase II/III studies in RR-ES if combination with IFOS is logical. rEECur is the first study to provide comparative toxicity and survival data for the four most commonly used chemotherapy regimens in RR-ES. Clinical trial information: ISRCTN36453794.