We evaluated intermittent and continuous thermodilution cardiac output data in 12 patients undergoing orthotopic liver transplantation. Measurements were performed at 16 predefined time points between induction of anaesthesia and 3 h after reperfusion of the liver graft. Cardiac output measurements yielded 192 data pairs (intermittent cardiac output range: 1.8-18.9 l.min-1, continuous cardiac output range: 3.3-20.0 l.min-1). During most of the procedure the correlation between intermittent and continuous cardiac output measurements was significant (r = 0.87, p < 0.0001), accompanied with a bias of -0.240 l.min-1 and a degree of precision of 1.789 l.min-1 (< 10.0 l.min-1:1.137 l.min-1, > or = 10.0 l.min-1:2.220 l.min-1). However, in the early phases after caval clamping and after reperfusion, accuracy was not acceptable. Only during these phases did the difference between the mean values of pulmonary artery blood temperature and rectal temperature increase (after caval clamping) or decrease (after reperfusion). In conclusion, despite acceptable levels of accuracy and precision between intermittent and continuous cardiac output measurement under stable conditions, both methods showed markedly decreased accuracy and precision in the early phases after caval clamping and after reperfusion, possibly owing to increased thermal noise.