We have previously developed a method for quantifying motion in carotid artery plaques from sequences of ultrasound (US) radiofrequency images. Here, we examine the intraoperator reproducibility of the results. Five independent recordings were made on each of six symptomatic and six asymptomatic patients, and processed off-line into 29 motion parameters, representing motion amplitude, stretch/compression and shear motion. For the statistical analysis, we used a linear mixed model and investigated the parameters for contributions from individual patients, contributions from recordings on each patient and contributions from heart cycles within each recording. The model was valid for seven parameters calculated over the entire heart cycle (four calculated over the systole only), which all showed good reproducibility (intraclass coefficient for variance over all patients rho(alpha) >/= 0.4). Averaging three recordings of two heart cycles each gives acceptable accuracy (normalised variance of patient means lambda < 0.3). This acquisition scheme is reasonable in a clinical situation.