Continuous measurement of cardiac output (CCO) is useful in assessing the cardiovascular status of patients during cardiac surgery and in intensive care. Recently, a CCO system (truCCOMS, Aortech, UK), capable of detecting rapid changes in cardiac output (CO) was introduced. The method is based on the energy required to maintain an integral heat-transfer device at constant temperature above the ambient value. The aim of this study was to assess the performance of this CCO system in vitro under in steady as well as pulsatile flow conditions representative of those in the pulmonary artery. In order to determine the sensitivity of the system to changes in vessel cross-sectional area and therefore local flow velocity, the catheter was deployed in a linear-tapered tube. Steady and pulsatile flows were generated, and the electrical power at various locations along the tapered tube was recorded. The results show significant differences in the performance under the two different flow conditions. In steady flow, the CO was highly dependent on the local velocity whereas in pulsatile flow, CO varied much less with local velocity. The sensitivity expressed as a percentage increase in CO per 100% increase in velocity at a CO of 5 l min(-1) was 87% in steady flow and 24% in pulsatile flow. Experiments carried out with three fluids with different viscosity show that the errors in determining CO in the tapered tube were also dependent on the Reynolds number and flow regime. The mean errors ranged from about 50% at 2 l min(-1) to less than 10% at 8 l min(-1). The correlation between the predicted and actual CO was generally good. In conclusion, the pulmonary artery catheter is not recommended in situations where blood flow is expected to be steady or of low pulsatility. It may, however, be suitable under normal pulsatile flow conditions in the pulmonary artery.