Transesophageal echocardiography (TEE) is increasingly used intraoperatively as a monitor of ventricular function and volume. Despite its increasing use, whether data from TEE monitoring can be interpreted accurately on-line in real-time is unknown. We studied the performance of five community-based, full-time cardiac anesthesiologists during 75 surgical procedures in which biplane TEE monitoring was used. Every 10 min intraoperatively, each anesthesiologist evaluated the video cine loop display of echocardiographic images to provide a real-time visual estimate of left ventricular ejection fraction area (EFA) and left ventricular filling at the level of the short axis and to assess regional wall-motion of the short axis and transgastric longitudinal views using a predefined scoring system. The same video images were analyzed quantitatively off-line by two blinded investigators. Intraoperative real-time estimates of EFA correlated moderately with off-line quantification (r = 0.8, P = 0.0001). Of the 662 cine loops analyzed by both off-line and real-time techniques, 386 (55%) were within +/-5% of each other, 495 (75%) were within +/-10% of each other, 561 (85%) were within +/-15% of each other, and 617 (93%) were within +/-20% of each other. The overall sensitivity and specificity of real-time echocardiographic ischemia detection were both 76%. However, there was individual variation among the five anesthesiologists. Recognition of normal and severe regional wall-motion abnormality, such as akinesis, had more concordance between real-time and off-line analysis, 93% and 79%, respectively, than recognition of mild regional wall-motion abnormalities. Anesthesiologists can estimate EFA in real-time to within +/-10% of off-line values in 75% of all cases. Real-time identification of normal regional function is more accurate than identification of abnormal function, i.e., there is variability in quantifying the severity of regional dysfunction.