A simplified thermodilution technique for cardiac output measurement is compared with the Fick method in dogs, over a wide range of cardiac output values. The reproducibility of the method is satisfactory (differences less than 10 per cent for 95 per cent of the measurements). The overall correlation factor (r = 0.97) is highly significant for cardiac outputs ranging from 750 to 7,500 ml.mn(-1). The influence of hematocrit (when lower than 30%) and of the injection site of indicator are discussed. Present experiments confirm that the use of pre-set factors allow for the thermal loss inside the catheter does not obviate the accuracy of thermodilution method for cardiac output measurements. The slope of the regression line is very close to 1. Lack of fit to linearity is tested and is found not significant throughout the explored range. This technique should therefore have extensive clinical applications.