OBJECTIVE The interchangeability of continuous measurement of cardiac output (CO) with the traditional bolus method in patients after cardiopulmonary bypass (CPB) is uncertain. METHODS Prospective observational clinical study. METHODS A 20-bed surgical ICU at a university hospital. METHODS Fourteen deeply sedated, ventilated, post-cardiac surgery patients, all equipped with a pulmonary artery catheter. METHODS Six hours after the end of the CPB, 56 simultaneous bolus and continuous measurements were compared by a linear regression analysis and Bland-Altman analysis. Bolus CO was estimated by averaging triplicate injections of 10 ml room-temperature NaCl 0.9%, delivered randomly during the respiratory cycle. A stringent maximum difference of 0.55 l min(-1) (about 10% of the mean bolus measured) was considered as a clinically acceptable agreement between the two types of measurements. To be interchangeable the limits of agreement (+/-2 SD of the mean difference between the two methods) should not exceed the chosen acceptable difference. RESULTS Continuous was correlated with bolus CO, with a correlation coefficient of r(2)=0.68. (p<0.01). The Bland-Altman analysis demonstrated an objective mean bias of 0.33+/-0.6 l min(-1) (confidence interval of -0.87-1.58) with 34% of measured values falling outside of the clinically acceptable limits. CONCLUSIONS Our results suggest that, in the first 6 h after CPB, continuous and bolus CO determinations are not interchangeable; one third of the values obtained by continuous CO fell outside the strict limits of clinically useful precision.