Six commercially available thermodilution cardiac output catheter-and-computer systems were evaluated over a variety of simulated clinical conditions using a fixed-output pulsatile-flow pumping unit. Simultaneous testing was performed at pumping rates of 60, 80, and 120 beats per minute and over an output range of 1.57 liters/min to 14.11 liters/min under normothermic, hyperthermic, and hypothermic conditions. Precision within each thermodilution system, as measured by the standard deviation (S.D.) of serial determinations, was acceptable with an overall S.D. range between 0.015 and 0.588 liters/min. When averaged over all experimental situations, the mean percent deviation of five or ten serial measurements was within 10 percent of the known pump output for all systems and maintained a fairly constant relationship above or below the pump output. In this in vitro test, iced injectate did not consistently improve precision or accuracy above that obtained with room-temperature injectate. Features other than performance should also be considered when selecting a particular unit to fulfill a clinical or research need.