Several predictive models exist for estimating the postoperative risks of abdominal aortic aneurysm (AAA) repair, although no particular tool has seen widespread use. We present the results of a multicenter, historic cohort study comparing the predictive capacity of the psoas muscle area (PMA), radiodensity (PMD), and lean muscle area (LMA) as surrogate markers of sarcopenia, over short- and long-term outcomes after AAA repair, compared to the mFI-5 and American Society of Anesthesiologists (ASA) scales. Retrospective review was conducted of all consecutive AAA elective repair cases (open or endovascular) in three tertiary-care centers from 2014 to 2019. Cross-sectional PMA, PMD, and LMA at the mid-body of the L3 vertebra were measured by two independent operators in the preoperative computed tomography. Receiver operating characteristic (ROC) curves were used to determine optimal cutoff values. Bivariate analysis, logistic regression, and Cox's proportional hazards models were built to examine the relationship between baseline variables and postoperative mortality, long-term mortality, and complications. 596 patients were included (mean age 72.7 ± 8years, 95.1% male, 66.9% EVAR). Perioperative mortality was 2.3% (EVAR 1.2% vs open repair 4.6%, p = .015), and no independent predictors could be identified in the multivariate analysis. Conversely, an age over 74years old (OR 1.84 95%CI 1.25-2.70), previous heart diseases (OR 1.62 95%CI 1.13-2.32), diabetes mellitus (OR 1.61 95%CI 1.13-2.32), and a PMD value over 66 HU (OR 0.58 95%CI 0.39-0.84) acted as independent predictors of long-term mortality in the Cox's proportional hazards model. Heart diseases (congestive heart failure or coronary artery disease), serum creatinine levels over 1.05mg/dL, and an aneurysm diameter over 60mm were independent predictors of major complications. Surrogate markers of sarcopenia had a poor predictive profile for postoperative mortality after AAA repair in our sample. However, PMD stood out as an independent predictor of long-term mortality. This finding can guide future research and should be confirmed in larger datasets.