Thoracocardiography noninvasively estimates changes in cardiac output by recording ventricular volume curves from an inductive plethysmographic transducer placed around the chest near the xiphoid process. We evaluated performance of thoracocardiography for estimation of cardiac output in 21 critically ill patients in comparison to thermodilution. A total of 201 paired cardiac output measurements were obtained over periods of 35 to 254 min. Since thoracocardiography tracks relative changes in cardiac output but does not provide absolute values, the first cardiac output by thermodilution in each patient was used to calibrate thoracocardiography for comparisons of subsequent cardiac output estimates to thermodilution. The mean difference (bias) of cardiac output (thoracocardiography - thermodilution) was 0.0 L/min, the limits of agreement (bias +/- 2 SD) included a range from -1.5 to +1.6 L/min. For estimations of relative changes in cardiac output by thoracocardiography and thermodilution the bias was 0%, and the limits of agreement -21 and +22%. We conclude that thoracocardiography is a promising noninvasive technique for monitoring cardiac output in critically ill patients.