The aim of this study was to determine if computed tomography (CT) psoas muscular attenuation measurements may predict all-cause mortality in patients undergoing TAVI. Ninety-four consecutive patients undergoing TAVI were analysed. The CT axial slice at the level of the fourth lumbar vertebra was selected. The psoas muscle areas were manually contoured. The circumferential surface area (CSA) of both psoas muscles was determined by selecting the voxels with attenuation values, ranging from 0 to 100 Hounsfield Units (HU). The mean CT attenuation coefficient of the psoas muscle (Psoas mean HU) was measured. The muscle was subdivided into a low-density muscle (LDM) (0-29 HU) and high-density muscle (HDM) (30-100 HU) portion. The HDM/LDM ratio was calculated. We searched for a correlation between HDM/LDM, CSA LDM (%), Psoas mean HU and all-cause mortality. The mean age was 81.2 ± 7.5 years. Thirty patients had adverse outcome (all-cause mortality). Compared with patients with the lowest CSA LDM (%), patients in the third and second tertiles had an increased hazard ratio for mortality (2.871; 95% confidence interval 0.880-9.371 and 5.044; 95% confidence interval 1.641-15.795, respectively) in a multivariable model with EuroSCORE II, Barthel frailty index and CSA LDM (%) (P = 0.231, 0.097 and 0.019, respectively). HDM/LDM and Psoas mean HU (as continuous variable) were also independent predictors of all-cause mortality (P = 0.019, P = 0.013, respectively). CSA LDM (%), Psoas mean HU and HDM/LDM are independent and incremental predictors of all-cause mortality in patients undergoing TAVI.