To evaluate a liver fibrosis quantitatively using an ultrasound B-mode image, it is effective to focus on a probability density function (PDF) of echo amplitudes. In our previous study, a multi-Rayleigh model which is a combination model of Rayleigh distributions with different variances was proposed to express the PDF of echo amplitudes from a fibrotic liver. Using the multi-Rayleigh model, the ultrasound B-mode image can be converted to the fibrotic probability image. To evaluate liver fibrosis quantitatively using the multi-Rayleigh model, a selection of a region of interest (ROI) is important because data such as an abdominal wall and a blood vessel wall are erroneously judged as the fibrotic tissue in the multi-Rayleigh model. In this paper, an automatic selection method of the ROI was examined and ultrasound B-mode images of the fibrotic liver were evaluated using the fibrotic probability images. The ROI could be selected automatically based on deviations of the PDF of echo amplitudes from the Rayleigh distribution and the multi-Rayleigh model. In the evaluation of fibrotic probability images, the averaged fibrotic probability became large as the progress of liver fibrosis. It was concluded that the progress of liver fibrosis could be evaluated quantitatively using fibrotic probability image based on the multi-Rayleigh model.