To provide useful information for diagnosis of atherosclerosis in addition to the imaging of morphology using the B-mode ultrasonography, we have developed a method in which an elasticity image is classified into tissue components using the reference data obtained by in vitro experiments. We have already measured the elasticity distributions for lipids, blood clots, fibrous tissue, and calcified tissue [H. Kanai, et al.: Circulation, 107, (2003) 3018, J. Inagaki, et al.: Jpn. J. Appl. Phys., 44, (2005) 4593]. From these previous studies, it was found that arterial tissues can be classified into soft tissues (lipids, blood clots) and hard tissues (fibrous tissue, calcified tissue) on the basis of their elasticity. However, it was difficult to differentiate lipids from blood clots and fibrous tissue from calcified tissue. Therefore, we proposed a tissue classification method using the likelihood function [J. Inagaki, et al.: Jpn. J. Appl. Phys., 45, (2006) 4732]. In this method, the elasticity distribution of each small region of interest (not a single pixel) in an elasticity image was used in classification of lipids, blood clots, fibrous tissue, and calcified tissue. In this paper, the optimum size of the region of interest was investigated to improve the tissue classification.