The study was undertaken to determine if the internal carotid artery peak systolic velocities (ICA PSVs) measured by two colour Doppler imaging systems (Acuson 128 and Siemens Quantum) agree sufficiently for the two systems to be interchangeable in evaluating carotid artery disease. One operator obtained blinded measurements of ICA PSV in 63 prospective nonrandomized patients at risk of stroke. The operator examined 20 patients in the first cohort to assess the intraobserver variation, and 43 patients in the second cohort to assess the limits of agreement between the systems. In vitro comparisons of the systems were also undertaken, using both string and flow phantoms. Excluding one outlier, the intraobserver reproducibility coefficient for both machines was 0.48 m s-1. The limits of agreement (within which 95% of differences lie) between systems were -0.47 to 0.45 m s-1. This reduced to -0.39 to 0.33 m s-1 when the one outlier was excluded. This is within the intraobserver reproducibility range. In vitro data show little intersystem variation with phantom velocity. Intratransducer differences increase when the Doppler angle is increased using the string phantom; maximum differences: Acuson 0.30 m s-1 (42%) and Siemens 0.32 m s-1 (32%). These are within the in vivo reproducibility range. Intratransducer difference decreases when the Doppler angle is increased using the flow phantom, maximum differences: Acuson 0.05 m s-1 and Siemens 0.07 m s-1. The results show the systems agree sufficiently to be interchangeable in evaluating carotid artery disease; however, errors in maximum PSVs, caused by operator or system variation, may lead to errors in percent stenosis grading of the carotid arteries.