Background: Five-year survival of RR-ES is about 15%. Several chemotherapy regimens are used, but without robust evidence. rEECur, the first randomised controlled trial in this setting, is defining a standard of care, balancing efficacy and toxicity. Methods: Patients aged 4 to 50 with RR-ES and fit to receive chemotherapy were randomised between topotecan & cyclophosphamide (TC), irinotecan & temolozomide (IT), gemcitabine & docetaxel (GD) or high-dose ifosfamide (IFOS). Primary outcome measure was objective response (OR) after 4 cycles by RECIST 1.1. Secondary outcomes included PFS, OS and toxicity. A probability-based Bayesian approach was used with multiple pairwise comparisons. At the first interim analysis patients allocated to GD had worse OR and PFS than the other arms and accrual to the GD arm was halted. The second interim assessment was planned to determine which arm should be closed when at least 75 evaluable patients had been recruited to the remaining arms and evaluated for the primary outcome measure. Results: 366 patients (87% RECIST-evaluable), recruited between 18/12/14 and 17/12/19, were randomised to TC (n=124), IT (118), GD (72) and IFOS (53). Median age was 20 years (range 4-49). Patients had: refractory disease (19%), first recurrence (66%), > first recurrence (14%). Initial disease site was bone in (66%). Sites of progression were: primary site only (16%) pleuropulmonary only (32%), other metastatic (52%). At median follow up of 9.2 months, outcome in the IT arm was: response rate 20%, median PFS 4.7 months (95% CI: 3.4 to 5.7), median OS 13.9 months (95% CI: 10.6 to 18.1). The table shows, for each pairwise comparison of IT with the other open arms (randomly labelled A and B to maintain blinding), the probabilities that OR, PFS and OS were better for X than for each other arm (RR = risk ratio, HR = hazard ratio). For OR, PFS and OS, all comparisons favoured arms A and B. The main grade 3/4 adverse events (% patients with an event) for IT (left hand values) compared with A and B pooled were: vomiting (6% v 1%), nausea (6% v 0%), diarrhoea (17% v 0%), fatigue (3% v 1%) and febrile neutropenia (3% v 24%). Conclusions: The first randomised trial in RR-ES has shown that IT, used as a control arm in planned and ongoing randomised phase II studies in RR-ES, is less effective than A and B in achieving tumour shrinkage or prolonging PFS and OS. The remaining two arms are continuing to recruit patients. Clinical trial information: ISRCTN36453794.