Application of the Doppler principle can provide relatively noninvasive and continuous measurement of cardiac output. However, it is based on certain assumptions that may introduce error. Esophageal Doppler cardiac output was compared with Fick cardiac output during acute blood loss (35-45% estimated blood volume) in eight anesthetized pigs. Mean Fick cardiac output decreased from 4.8 to 1.9 l/min, mean Doppler cardiac output from 4.9 to 2.9 l/min. This was accompanied by a decrease in mean arterial pressure from 119 to 55 mmHg and increase in heart rate from a mean of 115 to 156 beats/min. There was an inconsistent association between the two methods both within and between individual animals. Cubic polynomial regression equations of cardiac output with time indicated small measurement error in Fick (R2: mean 0.93, range 0.99-0.75) as opposed to Doppler (R2: mean 0.67, range 0.93-0.16) cardiac output. In one animal Doppler cardiac output showed an increase with time and in one the Doppler cardiac output measurements were unrelated to time. There was highly variable association comparing Fick versus Doppler cardiac output with correlations ranging from -0.76 to 0.98. A sign test for mean differences indicated that Doppler derived cardiac output was higher than Fick cardiac output, and the chance of this occurring if the true difference was zero was less than 1 in 1,000. A test for homogeneity of correlations was also rejected. Inaccuracies in individual assumptions in the computation of esophageal Doppler cardiac output, especially unaccounted changes in aortic diameter, are responsible for the inconsistent and unpredictable values of Doppler cardiac output obtained in this experimental model of hemorrhage.