We studied 21 adult cardiac surgical patients to assess the accuracy of continuous thermodilution method in comparison with bolus cardiac output measurement using room temperature versus cold injectates. An 8-French gauge thermal filament-wrapped, flow directed, pulmonary artery catheter was placed and connected to a computer system to measure continuous cardiac output(CCO) automatically. Both CCO and bolus cardiac output(BCO) measurements were performed at stable conditions after induction of anesthesia and also after weaning from cardiopulmonary bypass in each patient. As BCO measurement, 10 ml of ice-cold solution (BCO-IC) or room-temperature solution (BCO-RT) were injected using a closed delivery system with in-line temperature measurement. A total of 41 measurements were carried out. BCO-IC ranged from 2.2 to 9.7 L/min, BCO-RT from 2.1 to 9.9 L/min, and CCO from 2.3 to 12.4 L/min. Regression analysis demonstrated a close relationship between either two of three methods; correlation coefficients of [BCO-IC−BCO-RT], [BCO-IC−CCO], and [BCO-RT−CCO] were 0.986, 0.962 and 0.962, respectively. By the Bland and Altman analysis, biases (mean differences) were negligible among the three measurements(0.10-0.38 L/min), although precisions (standard deviation of differences) between [BCO-IC−CCO](0.59 L/min) and [BCO-RT−CCO](0.59 L/min) were slightly larger than that between [BCO-IC−BCO-RT](0.31 L/min). In conclusion, CCO methods are practically acceptable during cardiac surgery, and BCO-RT is equivalent in accuracy with BCO-IC, suggesting that BCO-RT can be used instead of BCO-IC because it is more convenient and less expensive.