Cardiac output measurement is part of routine monitoring in critically ill patients. In patients on extracorporeal lung assist, thermodilution cardiac output measurement may lead to erroneous results caused by indicator loss into the extracorporeal circuit. Seven patients on venovenous extracorporeal lung assist were studied using different extracorporeal blood flows. We compared conventional thermodilution cardiac output determinations with dye dilution cardiac output measurement, with dye injection into the pulmonary artery. The latter method is not affected by the extracorporeal circuit. The conventional thermodilution method overestimated cardiac output up to a maximum of 300%, providing results up to 10 L/min higher than true cardiac output. The mean difference between thermodilution and true cardiac output as determined by dye dilution with pulmonary artery indicator injection was 3.0 +/- 2.41 L/min. There was no correlation between thermodilution cardiac output values and true cardiac output (r = 0.06). We conclude that conventional thermodilution is not a suitable method for cardiac output measurement in patients on extracorporeal lung assist, especially if high extracorporeal blood flows are applied.