A thermodilution technique using a dual thermistor catheter introduced in the pulmonary artery was used for the determination of cardiac output at rest in 12 patients and for the same measurement during exercise in 6 patients with valvular disease. The correlation coefficient between thermodilution and the dye dilution technique is 0.98. The precision of the method assessed by multiple repeated determinations is 7.5% at rest and 6.9% during mild exercise. The reproducibility at rest indicates that a steady state has been maintained in an acceptable way. In one case of tricuspid insufficiency it was not possible to evaluate the cardiac output.