The Kubicek thoracic cylinder model of impedance cardiography (IC) for measuring beat-by-beat stroke volume (SV) was evaluated in controlled studies using the electromagnetic flowmeter (FM) as the reference technique. Assuming the validity of the Kubicek equation for stroke volume calculation, IC stroke volume was found to be a linear function of EM values at any one haematocrit over a wide range of SV, but the slope of the relationship fell as haematocrit fell. Experiments using the same equation in dogs, in which blood resistivity in vivo (ρ(τ)) was made the dependent variable, and the EM-derived value was used for stroke volume, showed that ρ(τ) was almost constant over a wide range of haematocrits. These findings were supported by studies in man and rabbit where Fick and thermodilution-derived values were used for stroke volume. When these data were applied to normotensive and hypertensive human subjects with normal hearts and lungs in controlled studies at rest, during tilting, with drug therapy and on exercise, IC measured stroke volume and cardiac output with a variability at least as good as the 9-11% acceptable for clinical use. This conclusion applied to thoracic configurations of different sizes and shapes from adult man to the neonate. In chronic disease states, while assessments of relative changes are valuable, absolute data are questionable. Further research is required under these conditions, as it is also for other models of IC, which are based on different assumptions.