Transcatheter aortic valve replacement (TAVR) is the preferred treatment option for older patients with symptomatic severe aortic stenosis. Differences in the properties of available TAVR systems can affect clinical outcomes. Among patients undergoing TAVR, we compared the self-expanding ACURATE neo TAVR system with the balloon-expandable SAPIEN 3 TAVR system with regard to early safety and efficacy. In this randomised non-inferiority trial, patients (aged ≥75 years) undergoing transfemoral TAVR for treatment of symptomatic severe aortic stenosis, and who were deemed to be at increased surgical risk, were recruited at 20 tertiary heart valve centres in Germany, the Netherlands, Switzerland, and the UK. Participants were randomly assigned (1:1) to receive treatment with the ACURATE neo or the SAPIEN 3 with a computer-based randomly permuted block scheme, stratified by study centre and Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM) category. The primary composite safety and efficacy endpoint comprised all-cause death, any stroke, life-threatening or disabling bleeding, major vascular complications, coronary artery obstruction requiring intervention, acute kidney injury (stage 2 or 3), rehospitalisation for valve-related symptoms or congestive heart failure, valve-related dysfunction requiring repeat procedure, moderate or severe prosthetic valve regurgitation, or prosthetic valve stenosis within 30 days of the procedure. Endpoint assessors were masked to treatment allocation. Non-inferiority of ACURATE neo compared with SAPIEN 3 was assessed in the intention-to-treat population on the basis of a risk-difference margin of 7·7% for the primary composite endpoint, with a one-sided α of 0·05. This trial is registered with ClinicalTrials.gov (number NCT03011346) and is ongoing but not recruiting. Between Feb 8, 2017, and Feb 2, 2019, up to 5132 patients were screened and 739 (mean age 82·8 years [SD 4·1]; median STS-PROM score 3·5% [IQR 2·6-5·0]) were enrolled. 30-day follow-up was available for 367 (99%) of 372 patients allocated to the ACURATE neo group, and 364 (99%) of 367 allocated to the SAPIEN 3 group. Within 30 days, the primary endpoint occurred in 87 (24%) patients in the ACURATE neo and in 60 (16%) in the SAPIEN 3 group; thus, non-inferiority of the ACURATE neo was not met (absolute risk difference 7·1% [upper 95% confidence limit 12·0%], p=0·42). Secondary analysis of the primary endpoint suggested superiority of the SAPIEN 3 device over the ACURATE neo device (95% CI for risk difference -1·3 to -12·9, p=0·0156). The ACURATE neo and SAPIEN 3 groups did not differ in incidence of all-cause death (nine patients [2%] vs three [1%]) and stroke (seven [2%] vs 11 [3%]); whereas acute kidney injury (11 [3%] vs three [1%]) and moderate or severe prosthetic aortic regurgitation (34 [9%] vs ten [3%]) were more common in the ACURATE neo group. TAVR with the self-expanding ACURATE neo did not meet non-inferiority compared to the balloon-expandable SAPIEN 3 device in terms of early safety and clinical efficacy outcomes. An early composite safety and efficacy endpoint was useful in discriminating the performance of different TAVR systems. Boston Scientific (USA).