The truCCOMS cardiac output monitor system provides a continuous and instantaneous measurement of cardiac output, derived from the amount of energy required for heating a filament to maintain a fixed 2 degrees C blood temperature difference between two thermistors located distally on a pulmonary artery catheter. Clinical studies, however, reported relatively poor accuracy of the cardiac output estimation, possibly due to linearly assumed power-cardiac output relationship used for calibration of the catheters. We experimentally studied the shape of the truCCOMS calibration relationship (i) in a hydraulic bench model of the right heart and (ii) in vivo intact animal model. The results showed a nonlinear relationship between the power input into the heating element and the cardiac output; which could satisfactorily be described with an exponential relationship. Comparison of the performance of the same catheters in vitro and in vivo showed that the in vitro determined calibration relationship should not be used for in vivo measurements. Finally, we also simulated the working principle of the catheter using a simplified numerical model of the blood flow and heat transfer around the catheter. The computed results also suggested a pronounced nonlinear relationship between power and cardiac output in pulsatile conditions. We conclude that the observed over- and underestimation of high- and low flows, respectively, by the current truCCOMS system is likely to arise from its linear calibration relationship. An appropriate calibration scheme accounting for the intrinsic nonlinear power-cardiac output relationship and the difference between in vitro and in vivo conditions should improve the clinical performance of the system.