BACKGROUND: A reliable non-invasive cardiac output measurement could enhance patient safety and reduce risk. Partial rebreathing cardiac output (RBCO) measurement is a non-invasive method based on a differential form of the CO2 Fick equation. The relative change in CO2 and ETCO2 in response to addition of dead space to the breathing circuit is used to measure cardiac output. The aim of this study was to compare this method in coronary artery bypass graft (CABG) patients during pre-cardiopulmonary bypass (CPB) and post-CPB with the currently accepted technique of intermittent thermodilution cardiac output (TDCO) measurement. METHODS: Eleven patients (n = 11, age = 50 +/- 13 years) undergoing CABG operations with CPB were studied. We measured the cardiac output non-invasively by using a RBCO monitor (NICO(R) , Novametrix Medical Systems Inc., USA) and used SpO2 and FiO2 to correct for intrapulmonary shunts. Invasively, using a pulmonary artery catheter (Hands-Off Thermodilution Catheter(R) , Arrow Co., USA) and 10 ml of iced 5% dextrose, an average of three consecutive TDCO was measured during end-expiration and compared with corresponding RBCO during pre-CPB and post-CPB for a total of 6 times respectively. Data was analyzed by paired T-test with significance set at P value < 0.05. RESULTS: Pre-CPB paired T-test analysis revealed no significant difference between partial rebreathing and bolus thermodilution cardiac output measurements. On the other hand, post-CPB differences between the two methods were significant (P < 0.05) and tended to decrease with time. Similarly, Pa-ETCO2 was increased abnormally after CPB, then decreased with time to a normal value. CONCLUSION: Post-CPB, partial rebreathing cardiac output did not correlate well with the thermodilution cardiac output. As a cause for the differences of the two cardiac output measurements, we couldconsider a thermal noise during thermodilution and an inadequate correction for the shunts in partial rebreathing measurements, but further investigation is needed.