Background Transcatheter aortic valve replacement (TAVR) is an effective therapeutic option for patients with severe aortic stenosis at high risk for surgery. Identification of causes of death after TAVR may help improve patient selection and outcome. Methods We enrolled 874 consecutive patients who underwent TAVR at 3 centers using all approved bioprostheses and different access routes. Clinical outcomes during follow-up were defined according to the Valve Academic Research Consortium 2 definitions. Causes of deaths were carefully investigated. Results Mean logistic European System for Cardiac Operative Risk Evaluation was 23.5% 15.3%; Society of Thoracic Surgery score, 9.0% +/- 8.2%. The Corevalve (Medtronic, Minneapolis, MN) was used in 41.3%; the Edwards Sapien (Edwards Lifesciences Inc., Irvine, CA) in 57.3%. Vascular access was transfemoral in 75.7%. In-hospital mortality was 5.0%. Cumulative mortality rates at 1 to 3 years were 12.4%, 23.4%, and 31.5%, respectively. Landmark analysis showed a significantly higher incidence of cardiovascular (CV) death in the first 6 months of follow-up and a significantly higher incidence of non-CV death thereafter. At Cox regression analysis, the independent predictors of in-hospital mortality were acute kidney injury grades 2 to 3 (hazard ratio [HR] 3.41) life-threatening bleeding (HR 4.26), major bleeding (HR 4.61), and myocardial infarction (HR 3.89). The independent predictors of postdischarge mortality were chronic obstructive pulmonary disease (HR 1.48), left ventricular ejection fraction at discharge (HR 0.98), and glomerular filtration rate