Prosthesis–patient mismatch (PPM) is associated with worse outcomes following surgical aortic valve replacement (SAVR). PPM has been identified in a significant proportion of TAVR, particularly in patients with small aortic annuli. Our objective was to evaluate the hemodynamic performances of balloon-expandable (BE) (Sapiens 3TM) versus two different self-expandable (SE) (Evolut ProTM, Accurate NeoTM) TAVR devices in patients with small aortic annulus defined by a computed tomography aortic annulus area (AAA) between 330 and 440 mm2. We enrolled 131 consecutive patients corresponding to 76 Sapiens 3 23 mm (58.0%), 26 Evolut Pro (19.9%) and 29 Accurate Neo (22.1%). Mean age was 82.5 ± 7.06 years, 22.9% of patients were male and mean Euroscore was 4.0%. Mean AAA was 374 ± 27 mm2 for Sapiens 3, 383 ± 29 mm2 for Corevalve Evolut Pro and 389 ± 25 mm2 for Accurate Neo. BE devices were associated with significantly higher rates of PPM (39.5%) as compared to SE devices (15.4% for Corevalve Evolut Pro and 6.9% for Accurate Neo) (p < 0.0001). Paravalvular leaks ≥ 2/4 were more often observed in SE devices (15.4% for Corevalve Evolut Pro and 17.2% for Accurate Neo) than in BE devices (2.6%) (p = 0.007). In conclusion, SE TAVR devices did achieve better hemodynamic results despite higher rates of paravalvular leaks. Therefore, SE TAVI devices could be considered as first choice in small aortic anatomy.