A previous controlled trial of autologous haematopoietic stem-cell transplantation (HSCT) in patients with refractory Crohn's disease did not meet its primary endpoint and reported high toxicity. We aimed to assess the safety and efficacy of HSCT with an immune-ablative regimen of reduced intensity versus standard of care in this patient population. This open-label, multicentre, randomised controlled trial was conducted in nine National Health Service hospital trusts across the UK. Adults (aged 18-60 years) with active Crohn's disease on endoscopy (Simplified Endoscopic Score for Crohn's Disease [SES-CD] ulcer sub-score of ≥2) refractory to two or more classes of biological therapy, with no perianal or intra-abdominal sepsis or clinically significant comorbidity, were recruited. Participants were centrally randomly assigned (2:1) to either HSCT with a reduced dose of cyclophosphamide (intervention group) or standard care (control group). Randomisation was stratified by trial site by use of random permuted blocks of size 3 and 6. Patients in the intervention group underwent stem-cell mobilisation (cyclophosphamide 1 g/m2 with granulocyte colony-stimulating factor (G-CSF) 5 μg/kg) and stem-cell harvest (minimum 2·0 × 106 CD34+ cells per kg), before conditioning (fludarabine 125 mg/m2, cyclophosphamide 120 mg/kg, and rabbit anti-thymocyte globulin [thymoglobulin] 7·5 mg/kg in total) and subsequent stem-cell reinfusion supported by G-CSF. Patients in the control group continued any available conventional, biological, or nutritional therapy. The primary outcome was absence of endoscopic ulceration (SES-CD ulcer sub-score of 0) without surgery or death at week 48, analysed in the intention-to-treat population by central reading. This trial is registered with the ISRCTN registry, 17160440. Between Oct 18, 2018, and Nov 8, 2019, 49 patients were screened for eligibility, of whom 23 (47%) were randomly assigned: 13 (57%) to the intervention group and ten (43%) to the control group. In the intervention group, ten (77%) participants underwent HSCT and nine (69%) reached 48-week follow-up; in the control group, nine (90%) reached 48-week follow-up. The trial was halted in response to nine reported suspected unexpected serious adverse reactions in six (46%) patients in the intervention group, including renal failure due to proven thrombotic microangiopathy in three participants and one death due to pulmonary veno-occlusive disease. At week 48, absence of endoscopic ulceration without surgery or death was reported in three (43%) of seven participants in the intervention group and in none of six participants in the control group with available data. Serious adverse events were more frequent in the intervention group (38 in 13 [100%] patients) than in the control group (16 in four [40%] patients). A second patient in the intervention group died after week 48 of respiratory and renal failure. Although HSCT with an immune-ablative regimen of reduced intensity decreased endoscopic disease activity, significant adverse events deem this regimen unsuitable for future clinical use in patients with refractory Crohn's disease. Efficacy and Mechanism Evaluation Programme, a Medical Research Council and National Institute for Health Research partnership.