Abstract: This research was conducted to predict the intramuscular fat percentage in longissimus muscle of live Beijing Black pig using real-time ultrasound image. The loin muscle area across the 10th to 11th rib, body weight, backfat thickness, loin muscle deepness and two longitudinal realtime ultrasound images were collected across the 9th to 13th rib and 5 cm off-midline on live pigs from 382 Beijing Black pigs. Gray gradient, gray level cooccurrence matrix and wavelet transform parameters whinin a defined region(80×80 pixel region across the 10th to 11th rib) for each ultrasound image were obtained using image analysis software (Matlab). After slaughter, a slice of longissimus muscle from left carcass across the 10th to 11th rib was cut off immediately for determining the intramuscular fat percentage (IMF) by the petroleum ether extraction method. The model to predict longissimus muscle intramuscular fat percentage (PIMF) was developed using linear regression analysis with carcass longissimus muscle intramuscular fat percentage (IMF) as dependent variables and body weight, backfat thickness, loin muscle area, loin muscle deepness and image parameters as independent variables. One hundred and twelve Beijing Black pigs were anew chose for model validation by correlation analysis of real intramuscular fat percentage and predicting intramuscular fat. The result of regression analysis indicated that 9 independent variables containing backfat thickness, loin muscle area and 7 image parameters were significant (P<0.05) in last model. The coefficient of determination and root mean square error for the prediction model were 0.305 8 and 0.006 5. The correlation analysis showed that the Pearson Correlation Coefficients and Spearman Correlation Coefficients were 0.553 4 and 0.627 2 (P<0.000 1). The result indicated that using real-time ultrasound image to predict intramuscular fat percentage in live pig was feasible. And this method can be used in Beijing Black pigs’ breeding work and developing the intramuscular fat effectively.