Cardiac output (Q) was measured with the thermodilution technique at 4 points during the respiratory cycle in dogs. Boluses of an ice-cold solution were injected at mid- and end-inspiration while the animals were on intermittent positive-pressure ventilation (IPPV) or after induction of positive end-expiratory pressure (PEEP), with and without induction of experimental respiratory failure. Values were most constant at end-inspiration. During IPPV without respiratory failure, the standard error of 74 measurements at end-inspiration was 5.1% and of 74 measurements randomly selected was 9.8%. Continuous infusion resulted in similar fluctuations in Q (r = 0.92). Values obtained during experimental respiratory failure were not significantly different.The authors conclude that timing the bolus injections with the respiratory cycle enhances the reproducibility of Q values obtained with the thermodilution technique.