Abstract Aims The objective was to compare the incidence and impact on outcomes of measured (PPMM) vs. predicted (PPMP) prosthesis–patient mismatch following transcatheter aortic valve replacement (TAVR). Methods and results All consecutives patients who underwent TAVR between 2007 and 2018 were included. Effective orifice area (EOA) was measured by Doppler-echocardiography using the continuity equation and predicted according to the normal reference for each model and size of valve. PPM was defined using EOA indexed (EOAi) to body surface area as moderate if ≤0.85 cm2/m2 and severe if ≤ 0.65 cm2/m2 (respectively, ≤ 0.70 and ≤ 0.55 cm2/m2 if body mass index ≥ 30 kg/m2). The outcome endpoints were high residual gradient (≥20 mmHg) and the composite of cardiovascular mortality and hospital readmission for heart failure at 1 year. Overall, 1088 patients underwent a TAVR (55% male, age 79.1 ± 8.4 years, and STS score 6.6 ± 4.7%); balloon-expandable device was used in 83%. Incidence of moderate (10% vs. 27%) and severe (1% vs. 17%) PPM was markedly lower when defined by predicted vs. measured EOAi (P < 0.001). Balloon-expandable device implantation (OR: 1.90, P = 0.029) and valve-in-valve procedure (n = 118; OR: 3.21, P < 0.001) were the main factors associated with PPM occurrence. Compared with measured PPM, predicted PPM showed stronger association with high residual gradient. Severe measured or predicted PPM was not associated with clinical outcomes. Conclusion The utilization of the predicted EOAi reclassifies the majority of patients with PPM to no PPM following TAVR. Compared with measured PPM, predicted PPM had stronger association with haemodynamic outcomes, while both methods were not associated with clinical outcomes.