In patients on veno-arterial extracorporeal membrane oxygenation (ECMO) premembranous venous oxygen saturation (S_pm-vO₂) is continuously displayed on the ECMO console. However, the concordance between S_pm-vO₂ and mixed venous oxygen saturation (S_vO₂) remains largely unexplored. Our single-center retrospective study included adult patients who had paired S_vO₂ and S_pm-vO₂ readings within 15 minutes of each other, on peripherally cannulated V_{f ivc}-A ECMO and a pulmonary artery using catheter (PAC). The 82 pairs of observations showed a mean difference of 11.37% (95% limits of agreement −6.0 to 28.74, p < 0.001) between S_pm-vO₂ and S_vO₂. Although the two values correlated with each other (r = 0.51, p < 0.01), the difference between the paired measurements was larger at lower values of S_vO₂ (3.72 ± 6.38% when S_vO₂ >80%, 11.79±7.46% when S_vO₂ between 60% and 80%, and 18.81±12.09% when S_vO₂ <60%). The equation S_vO₂ = 1.2* S_pm-vO₂ – 28.03 was obtained by Passing Bablok regression. Cardiac index calculated by S_pm-vO₂ and S_vO₂ differed by 0.8L/minute/m² (95% limits of agreement −0.52 to 2.17, p < 0.001). In peripheral VA-ECMO, S_pm-vO₂ is consistently higher than S_vO₂, with more discordance at lower saturation levels. Using S_pm-vO₂ to estimate cardiac output using Fick method yields inaccurate results.