Transcatheter aortic valve replacement (TAVR) is established for selected patients with severe aortic stenosis. However, limitations such as suboptimal deployment, conduction disturbances, and paravalvular leak occur. To evaluate if a mechanically expanded valve (MEV) is noninferior to an approved self-expanding valve (SEV) in high-risk patients with aortic stenosis undergoing TAVR. The REPRISE III trial was conducted in 912 patients with high or extreme risk and severe, symptomatic aortic stenosis at 55 centers in North America, Europe, and Australia between September 22, 2014, and December 24, 2015, with final follow-up on March 8, 2017. Participants were randomized in a 2:1 ratio to receive either an MEV (n = 607) or an SEV (n = 305). The primary safety end point was the 30-day composite of all-cause mortality, stroke, life-threatening or major bleeding, stage 2/3 acute kidney injury, and major vascular complications tested for noninferiority (margin, 10.5%). The primary effectiveness end point was the 1-year composite of all-cause mortality, disabling stroke, and moderate or greater paravalvular leak tested for noninferiority (margin, 9.5%). If noninferiority criteria were met, the secondary end point of 1-year moderate or greater paravalvular leak was tested for superiority in the full analysis data set. Among 912 randomized patients (mean age, 82.8 [SD, 7.3] years; 463 [51%] women; predicted risk of mortality, 6.8%), 874 (96%) were evaluable at 1 year. The primary safety composite end point at 30 days occurred in 20.3% of MEV patients and 17.2% of SEV patients (difference, 3.1%; Farrington-Manning 97.5% CI, -∞ to 8.3%; P = .003 for noninferiority). At 1 year, the primary effectiveness composite end point occurred in 15.4% with the MEV and 25.5% with the SEV (difference, -10.1%; Farrington-Manning 97.5% CI, -∞ to -4.4%; P<.001 for noninferiority). The 1-year rates of moderate or severe paravalvular leak were 0.9% for the MEV and 6.8% for the SEV (difference, -6.1%; 95% CI, -9.6% to -2.6%; P < .001). The superiority analysis for primary effectiveness was statistically significant (difference, -10.2%; 95% CI, -16.3% to -4.0%; P < .001). The MEV had higher rates of new pacemaker implants (35.5% vs 19.6%; P < .001) and valve thrombosis (1.5% vs 0%) but lower rates of repeat procedures (0.2% vs 2.0%), valve-in-valve deployments (0% vs 3.7%), and valve malpositioning (0% vs 2.7%). Among high-risk patients with aortic stenosis, use of the MEV compared with the SEV did not result in inferior outcomes for the primary safety end point or the primary effectiveness end point. These findings suggest that the MEV may be a useful addition for TAVR in high-risk patients. ClinicalTrials.gov Identifier: NCT02202434.