We evaluate the accuracy of cardiac output measurement with esophageal Doppler ultrasonography (ECO). A total of 71 simultaneous measurement of esophageal Doppler and thermodilution cardiac output were compared in 16 patients undergoing general anesthesia in the supine position. ECO was determined easily with minimal experience, and significantly correlated with thermodilution cardiac output (TDCO) measurement ( P < 0.001). The regression equation obtained was Y = 0.983X + 0.019, and the correlation coefficient was 0.935. Furthermore, ECO was more reproducible than TDCO. However, ECO is not able to assess CO accurately in either lateral or prone position and after cardiopulmonary bypass in open heart surgery. Our results suggest that the esophageal Doppler technique allows a noninvasive and continuous cardiac output monitoring in patients during general anesthesia, and that it is more useful in patients for whom invasive monitoring is considered inappropriate. However, further improvement in this technique will be necessary for its routine use in clinical anesthesia.