Brachial artery ultrasonography is used to measure flow-mediated dilatation (FMD) as a marker of endothelial function in patients at risk for atherosclerosis. Major disadvantages are the time-consuming manual readings and the high within-and between-observer variability. The authors hypothesize that the ultrasound-based determination of endothelial function can be simplified and refined by an automated analysis system. FMD was quantified by a 7.5 MHz linear transducer following 5 minutes of ischemia of the proximal forearm in 8 healthy volunteers on two occasions. Brachial artery diameter was comparatively assessed 1. manually from the video signal and 2. by a PC-based analyzing system. For the manual readings the mean differences for the FMD were 2.5±2.3% between-reader, 2.0±0.9% within-reader and 2.1±1.5% for scans on different days in contrast to 0.8±0.4 (between-reader), 0.8±0.6 (within-reader) and 1.3±0.9%(day-to-day) for the computerized system. The coefficient of variability for the measurement of arterial diameter was 1.34% for manual readings and 0.78% for the automated analysis system. The mean time for manual readings from S-VHS tapes was 35 minutes in contrast to 9 minutes for the PC-based analysis system. The new automated analysis system for the boundary detection of the vascular wall reduces the variability and greatly increases the speed of the measurements to determine endothelial function. In future, these advantages will help to screen larger numbers of individuals for endothelial dysfunction, particularly for follow-up and intervention trials, and to reduce the variability between different laboratories.