AbstractThe thermal dilution technique and its suitability for automatic computation of left ventricular volumes were evaluated on 28 anaesthetized dogs. The end‐diastolic volume (EDV) was computed from ventricular washout curves in two ways, first utilizing the ratio between the indicator amount and its peak concentration (first ejection method); and secondly the ratio between stroke volume and ejection fraction with the latter obtained by subtracting the residual fraction from 1.0. The residual fraction was determined as the mean of ratios between successive steps of four ventricular ejections on the downslope of the curve (downslope method). The linear correlation coefficient between the EDV determinations (n = 366) was 0.926, and slope y = 1.14 × + 0.5 ml with 4.6 ml as standard error of estimate. The first ejection method gave on average 15% higher EDV values than the downslope method. When heart rate was increased by pacing from 88 to 240/min, both methods showed a reduction of the EDV, but the downslope method to a greater degree. When after load was increased but heart rate decreased by phenylephrine, the first ejection method but not the downslope method indicated an increase in EDV. Otherwise, when pronounced tachy‐ or bradycardia were avoided, ventricular volumes yielded by the two methods were comparable. Indicator injection did not substantially affect ventricular pressure dynamics. The first ejection method seems to be a simple means for repeated measurements of EDV even in the case of diseased he