The modified algorithm for the non-invasive determination of cardiac output (CO) by electrical bioimpedance-electrical velocimetry (EV)-has been reported to give reliable results in comparison with echocardiography and pulmonary arterial thermodilution (PA-TD) in patients either before or after cardiac surgery. The present study was designed to determine whether EV-CO measurements reflect intraindividual changes in CO during cardiac surgery. Prospective, observational study. Operating room (OR) and intensive care unit (ICU) of a university hospital. Twenty-nine patients undergoing elective cardiac surgery. None. CO was determined simultaneously by PA-TD and EV after induction of anesthesia (t1) and 4.9+/-3.5 h after ICU admission (t2). TD-CO was 3.9+/-1.4 and 5.4+/-1.1 l/min at t1 and t2 (p < 0.0001). EV-CO was 4.3+/-1.1 and 4.9+/-1.5 l/min at t1 and t2 (p = 0.013). Bland-Altman analysis showed a bias of -0.4 l/min and 0.4 l/min and a precision of 3.2 and 3.6 l/min (34.3% and 67.4%) at t1 and t2, respectively. Analysis of the individual pre- to postoperative changes in CO with both methods revealed bidirectional changes in n = 12 patients and unidirectional changes with a difference greater than 50% and less than 50% in n = 9 and n = 8 patients, respectively. The disagreement between PA-TD and EV-CO measurements after anesthesia induction and after ICU admission, as well as the fact that thoracic bioimpedance did not adequately reflect pre- to postoperative changes in CO, questions the reliability of EV-CO measurements in cardiac surgery patients and contrasts sharply with previous studies.