Two commercial automated, non-invasive systems for estimation of cardiac output were evaluated. Values of cardiac output obtained by electrical bioimpedance cardiography (BoMed NCCOM3 machine) were compared with values derived from an indirect Fick technique that uses carbon dioxide rebreathing (Gould 9000 IV system) during 103 simultaneous measurements made at rest in 19 randomly selected subjects and on exercise in 11 subjects. Cardiac output values obtained with impedance cardiography were significantly correlated with those measured by the indirect Fick method, although there was a wide scatter with over 73% of the readings lying outside the limits defined by the line of identity +/- 20%. This correlation was greatly reduced when stroke volume index was used instead of cardiac output. Indirect Fick results were linearly related to oxygen uptake both at rest and on exercise, while impedance cardiography results did not correlate with oxygen uptake. Impedance cardiography gave consistently lower results for cardiac output than indirect Fick at all levels of exercise. Both machines were easy to use and produced acceptable mean (SE) coefficients of variation (BoMed NCCOM3 7.7 (1.0)%, Gould 9000 IV 10.6 (1.4)%). Further validation is required before either of these machines can be recommended as an alternative to invasive monitoring in clinical practice.