Accurately estimating the channel distortion for the video communication is significant to the joint power and rate-distortion (P-R-D)optimization. This paper proposes a channel distortion estimation model based on MPEG-4 coding stream. The model recursively estimates the channel distortion of the video coding sequences according to the bit error rate of the channel and the video coding information obtained in the video encoder. Simulation results demonstrate that the average relative estimation error and the average absolute estimation error introduced by the model are small. The model is applicable for the relative optimizations based on P-R-D.