SummaryAfter cardiac surgery, patients with low left ventricular ejection fraction probably benefit the most from accurate monitoring of continuous cardiac output. Thirty patients with impaired ventricular function were studied, and intermittent bolus thermodilution and continuous pulse contour (LiDCO plusTM) cardiac output compared. Following lithium dilution calibration, a total of 220 paired results were recorded. Thermodilution and LiDCO measurements ranged from 2.3 to 11.0 and 2.6 to 10.8 l.min−1, respectively. Corresponding means (SD) were 6.1 (1.6) and 6.2 (1.9) l.min−1, with coefficients of variance of 26 and 31%, respectively. The correlation coefficient was 0.82, bias 0.28 l.min−1 with upper and lower limits of agreement 1.96 and −1.41 l.min−1; the percentage error was 27%. LiDCO showed good correlation, marginal bias and acceptable limits of agreement and percentage error. It could therefore potentially replace thermodilution as a means of measuring cardiac output in the ICU, particularly when determination of pulmonary artery pressure is not required.