Objective: Results from randomized controlled trials (RCTs) and real-world study (RWS) appear to be discordant. We aimed to investigate whether data derived from RCTs and RWS evaluating long-term all-cause mortality of transcatheter aortic valve implantation (TAVI) versus surgical aortic valve replacement (SAVR) in patients with severe aortic stenosis (AS) were in agreement. Methods: RCTs or RWS comparing TAVI and SAVR, reporting longterm (≥2-year follow-up) all-cause mortality, were identified. We also carried out subgroup analyses to access the effect in different subgroups. A pre-designated data extraction form including 5 domains and 26 items was used to explore the relationship between RCTs and RWS. Mortality and effect in different subgroups were evaluated using random-effects meta-analyses. Results: Five RCTs (5421 participants, TAVI: 2759, SAVR: 2662) and 33 RWS (20839 participants; TAVI: 6585, SAVR: 14254) were identified. Pooled RCT analysis showed no difference in all-cause mortality between TAVI and SAVR (HR=0.97, 95% CI: 0.88-1.07; P=0.55). In RWS, TAVI was associated with an increased risk of allcause mortality (HR=1.46, 95% CI: 1.26-1.69; P<0.001) compared to SAVR. Conclusion: These results highlight the inconsistencies between RCTs and RWS in assessing long-term all-cause mortality in the treatment of AS using TAVI or SAVR, which may be caused by interactions of clinical characteristics or study design. RCTs as well as RWS are both developing and improving; the advantages of one kind of design, measurement and evaluation can and should be thoughtfully referred to the other.