To evaluate the clinical utility of a new device for continuous noninvasive cardiac output monitoring (NICOM) based on chest bio-reactance compared with cardiac output measured semi-continuously by thermodilution using a pulmonary artery catheter (PAC-CCO). Prospective, single-center study. Intensive care unit. Consecutive adult patients immediately after cardiac surgery. Cardiac output measurements obtained from NICOM and thermodilution were simultaneously recorded minute by minute and compared in 110 patients. We evaluated the accuracy, precision, responsiveness, and reliability of NICOM for detecting cardiac output changes. Tolerance for each of these parameters was specified prospectively. A total of 65,888 pairs of cardiac output measurements were collected. Mean reference values for cardiac output ranged from 2.79 to 9.27 l/min. During periods of stable PAC-CCO (slope<+/-10%, 2SD/mean<20%), the correlation between NICOM and thermodilution was R=0.82; bias was +0.16+/-0.52 l/min (+4.0+/-11.3%), and relative error was 9.1%+/-7.8%. In 85% of patients the relative error was <20%. During periods of increasing output, slopes were similar with the two methods in 96% of patients and intra-class correlation was positive in 96%. Corresponding values during periods of decreasing output were 90% and 84%, respectively. Precision was always better with NICOM than with thermodilution. During hemodynamic challenges, changes were 3.1+/-3.8 min faster with NICOM (p<0.01) and amplitude of changes did not differ significantly. Finally, sensitivity of the NICOM for detecting significant directional changes was 93% and specificity was 93%. Cardiac output measured by NICOM had most often acceptable accuracy, precision, and responsiveness in a wide range of circulatory situations.