Thermodilution methods to determine cardiac output (CO) may be affected by veno-venous extracorporeal membrane oxygenation (ECMO). We compared CO estimations by pulmonary arterial thermodilution using a pulmonary arterial catheter (CO_PAC), transpulmonary thermodilution (CO_TPTD), and three-dimensional echocardiography (3DEcho) (CO_3DEcho) in 18 patients under veno-venous ECMO. Comparisons between CO_3DEcho and CO_PAC, and CO_TPTD were performed using correlation statistics and Bland–Altman analysis. Blood flow on ECMO support ranged from 4.3 to 5.8 L/min (median 4.9 L/min). Cardiac output measured with three-dimensional echocardiography was 5.2 L/min (3.8/5.9), CO_PAC was 7.3 L/min (5.9/7.9), and CO_TPTD was 7.3 L/min (6/8.2) (median [25%/75% percentile]). Bland–Altman analysis of CO_3DEcho and CO_PAC revealed a mean bias of −2.06 L/min, with limits of agreement from −4.16 to 0.04 L/min. Bland–Altman analysis of CO_3DEcho and CO_TPTD revealed a mean bias of −2.22 L/min, with limits of agreement from −4.18 to −0.25 L/min. We found a negative mean bias and negative limits of agreement between CO_3DEcho and CO_PAC/CO_TPTD. We concluded an influence on the estimation of CO by thermodilution under ECMO most likely due to loss of indicator resulting in an overestimation of CO. Clinicians should consider this when monitoring thermodilution-based CO under ECMO.