Data from 710 yearling bulls and steers collected from 1991 to 1994 were used to predict the percentage of intramuscular fat (PIFAT) by using real-time ultrasound (RTU) and imageprocessing parameters. Image-processing parameters included histogram, texture, and Fourier transformation parameters. Additionally, ultrasound fat thickness (UFAT) was included. Two multiple regression models Model1 excluding UFAT and Model2 including UFAT,were developed by using 392 images and validated with 318 independent images. These models were used to assess the accuracy of image parameters in predicting PIFAT and to determine whether including UFAT as an additional covariate parameter increases accuracy. Results indicated that for actual PIFAT values ranging from .5% to 13%, RTU and image-processing parameters can consistently predict PIFAT with a root mean square error (RMSE) of 1.43 and 1.41 and a coefficient of determination (R-square) of .59 and .6 for Model1 and Model2, respectively. Both models were unbiased with intercepts of .47 and .51 (p > 0.1), respectively. RTU and image-processing parameters can accurately and without bias predict PIFAT without including UFAT in the prediction model.