More than 900 real-time ultrasound images from ribeye muscles across the 11th to 13th ribs of live beef animals were collected over the period of four years. Fat has acoustic properties appreciably different from those of other soft tissues, causing the transmitted ultrasound to be reflected from the interfaces between fat and muscle. As fat deposits (and hence marbling) increase, the speckle content of B-mode images also increased. Because speckle alters the texture of the image, intramuscular %-fat (IMFAT) can be estimated using texture analysis. A region of interest (ROI) was selected from the acquired image subjectively so that it contained a good image of the ribeye area between and above the 12th and 13th ribs. Image-processing techniques were applied for extracting features to analyze textures. The features showing good correlations with the actual IMFAT were used to develop statistical classification models. Because overall accuracy of prediction was improved by developing different regression models for the low-IFAT (less than 8%) and high-IFAT (more than 8%) groups (refer to R1325), statistical pattern recognition and classification techniques were applied to pre-classify the images into the low- or high-IFAT groups. The classification tree provided overall classification accuracy of 90% for lowand high- IFAT groups of images.