Upper-body peripheral arterial return is used during veno-arterial extracorporeal membrane oxygenation (VA-ECMO), but its clinical tradeoffs versus femoral return are uncertain. We reviewed observational studies comparing upper-body to femoral arterial return for peripheral VA-ECMO in cardiogenic shock and related rescue states. Seven retrospective cohorts (2,987 patients) were included. Upper-body return included axillary, subclavian, and combined approaches with direct and graft-mediated techniques. In the primary short-term mortality analysis (five studies; 2,317 patients), mortality did not differ significantly between upper-body and femoral return (odds ratio [OR]: 1.01, 95% confidence interval [CI]: 0.62–1.65; p = 0.96; I² = 44%). In crude pooled analyses, upper-body return was associated with higher ischemic stroke (six studies; OR: 1.78, 95% CI: 1.31–2.42; p = 0.005; I² = 0%), hemorrhagic stroke (four studies; OR: 2.34, 95% CI: 1.41–3.87; p = 0.01; I² = 0%), and any stroke/cerebrovascular event (five studies; OR: 1.63, 95% CI: 1.28–2.09; p = 0.005; I² = 0%). Limb ischemia was extremely imprecise and not statistically significant (three studies; OR: 0.34, 95% CI: < 0.01–44.78; p = 0.44; I² = 82%). The available observational evidence does not suggest the superiority of either access strategy over the other. Given the heterogeneous data and stroke signal, the arterial return strategy should be individualized rather than broadly adopted as a default approach.