BACKGROUND: Impedance Cardiography is a noninvasive and simple method of cardiac output determination. The purpose of this study was to compare an impedance device with the thermodilution method during the intraoperative period in patients undergoing open heart surgery. METHODS: The study was undertaken in 12 patients undergoing elective open heart surgery without a shunt. When abnormal impedance signals were obtained, the patients were not included in the analysis. Stroke volume was calculated according to the formula of Bernstein. Cardiac output was measured simultaneously by the impedance cardiograph and thermodilution method. We obtained 41-paired measurements during the surgical procedure. The method described by Bland and Altman and linear regression analysis were used for comparison. RESULTS: The correlation coefficient between the two methods was r = 0.36 (P<0.05). Bias and precision analysis between the two techniques showed a mean difference between techniques of 0.80 L/min and a SD of the differences of 1.71 L/min (95% level of agreement 4.21 L/min to -2.62 L/min). Trending analysis showed the impedance cardiogram to inaccurately track the direction of thermodilution CO changes and to underestimate their magnitude (r = 0.57, intercept -3.29, slope 1.27). CONCLUSIONS: This study reveals clinically significant errors in impedance CO measurements. These inaccuracies may be related to the intraoperative environment and abnormal cardiovascular function. Further investigation of such techniques to improve the performance of the impedance cardiogram in the intraoperative setting is warranted.