The contribution of veno-venous (VV) extracorporeal membrane oxygenation (ECMO) to systemic oxygen delivery is determined by the ratio of total extracorporeal blood flow (Q˙_EC) to cardiac output (Q˙). Thermodilution-based measurements of Q˙ may be compromised by blood recirculating through the ECMO (recirculation fraction; Rf). We measured the effects of Q˙_EC and Rf on classic thermodilution-based measurements of Q˙ in six anesthetized pigs. An ultrasound flow probe measured total aortic blood flow (Q˙_A0) at the aortic root. Rf was quantified with the ultrasound dilution technique. Q˙_EC was set to 0–125% of Q˙_A0 and Q˙ was measured using a pulmonary artery catheter (PAC) in healthy and lung injured animals. PAC overestimated Q˙ (Q˙_Pa) at all Q˙_EC settings compared to Q˙_A0. The mean bias between both methods was 2.1 L/min in healthy animals and 2.7 L/min after lung injury. The difference between Q˙_Pa and Q˙_A0 increased with an Q˙_EC of 75–125%/Q˙_A0 compared to Q_EC <50%/Q˙_A0. Overestimation of Q˙_Pa was highest when Q˙_EC resulted in a high Rf. Thus, thermodilution-based measurements can overestimate cardiac output during VV ECMO. The degree of overestimation of Q˙_Pa depends on the Q˙_EC/Q˙_A0 ratio and the recirculation fraction.