A new continuous thermodilution cardiac output measurement technique and companion flow-directed pulmonary artery catheter were evaluated in intensive care unit (ICU) patients. Continuous cardiac output was monitored for 6 hours in each patient, and, at selected intervals, a series of bolus thermodilution cardiac output determinations was made and averaged for comparison. A total of 222 data pairs was obtained in 54 patients. The cardiac outputs ranged from 2.8 to 10.8 L/min. The linear regression is represented by the following equation: continuous thermodilution = 0.99 bolus thermodilution + 0.02. The correlation coefficient r was 0.94, the Syx was 0.54. The mean relative error was 0.3%, and the standard deviation of the relative error was 11.5%. The absolute measurement bias was 0.02 L, and the 95% confidence limits were 1.07 and -1.03 L. The results demonstrated that the new continuous thermodilution cardiac output measurement technique provided acceptable accuracy and was considerably easier to use in the clinical situations studied in the ICU.