The purpose of this work was to process carotid plaque ultrasound images, employing pattern recognition methods, for assessing the embolization risk factor associated with carotid plaque composition. Carotid plaques of 56 ultrasound images displaying carotid artery stenosis were categorized by means of the gray scale median (GSM) as high risk of causing brain infarcts (GSM≤50 gray level) and low-risk (GSM>50 gray level), and in accordance with the physician’s assessment and final clinical outcome. In each plaque image, the ratio of echo-dense to echo-lucent area was automatically calculated and it was combined with other textural features, calculated from the image histogram, the co-occurrence matrix, and the run-length matrix. These features were employed as input to two classifiers, the quadratic Bayesian (QB accuracy 91.7%) and the support vector machine (SVM accuracy 100%), which were trained to characterize plaques as either high risk or low risk of causing brain infarcts.