A method of classification of plaques using the gray level distribution of pixels in ultrasonic images of carotid plaques. The method and algorithm classifies plaques into 12 classes. Each class is associated with a different level of risk of developing symptoms.