Changes in thoracic electrical bioimpedance during the cardiac cycle are utilised by the BoMed NCCOM3 monitor to measure cardiac output (COTEB). The technique provides a continuous noninvasive measurement but it has not been widely accepted. To determine the accuracy of the monitor, we compared its measurement with cardiac output measured by dye dilution (CODD) during induced hypotension and recovery in 23 dogs. After calibration of the NCCOM3 monitor during a resting state in each dog [mean blood pressure 112 +/- 17 (SD), mean CODD 3.22 +/- 0.99 l/min], the mean difference (COTEB-CODD) between paired measurements at the nadir of hypotension (blood pressure 55 +/- 24 mmHg) was 0.29 +/- 0.47 l/min whose limits of agreement (mean difference +/- 2 SD) were + 111.8% and -59.1% of the mean hypotensive CODD (1.10 +/- 0.66 l/min). Upon recovery from hypotension (mean blood pressure 102 +/- 20 mmHg), the mean difference between paired measurements was -0.28 +/- 0.66 l/min, whose limits of agreement were +44.1% and -67.8% of the mean CODD (2.36 +/- 1.01 l/min). The mean difference between the two techniques is too variable and excessive to permit substitution of one technique for the other. These results do not support the accuracy and reliability of the BoMed NCCOM3 cardiac output monitor.