We evaluated a new, ultra-fast response continuous cardiac output monitor in 34 adult patients undergoing off-pump coronary artery bypass graft surgery. Cardiac output was measured with the TruCCOMS continuous cardiac output monitor (Aortech International plc, Lanarkshire, UK), using triplicate cold bolus thermodilution as the criterion standard, at fixed time points during surgery and during dobutamine infusion. The two techniques were compared using linear regression and Bland-Altman analysis. Overall, the study device displayed a bias of 0.4 l.min(-1) with limits of agreement of +2.5 l.min(-1) and -1.7 l.min(-1). The study device failed to detect the change in cardiac output caused by dobutamine accurately (y = 0.18x + 0.45; r(2) = 0.13), with an error linearly related to the magnitude of the change measured. We conclude that the device's failure to detect changes in cardiac output could be a major limitation in its clinical use in its current form.