This study examined the effect of impedance algorithm adjustment to reflect abnormalities found in cardiac output estimation in the intensive care unit. Impedance (Kubicek and Sramek equations) and thermodilution were measured concurrently in 61 patients. The mean difference between Kubicek and thermodilution (n=40) was 1.47 L/min (95% confidence interval [CI], 0.47–2.47) and between Sramek and thermodilution (n=54) was 2.68 L/min (95% CI, 1.93–3.44). Exclusion of patients with valve regurgitation improved agreement between Kubicek and thermodilution (n=32), with a mean difference of 2.02 L/min (95% CI, 1.10–2.94). Multiple regression determined the role of skinfold thickness, pH, hematocrit, sodium, chloride, albumin, protein, and urea within impedance. Kubicek was recalculated using the new algorithm and recompared with thermodilution. The mean difference was −0.38 L/min (95% CI, −1.92 to 1.16). This study found poor agreement between impedance and thermodilution in critically ill patients, but exclusion of those with valve regurgitation and adjustment for hematocrit and skinfold thickness improved agreement.