Abstract: This paper presents a novel multiple description video coding algorithm for H.264, called mode duplication based multiple description coding. Each frame in the input video sequence is down-sampled first horizontally and then vertically to form four sub-frames. The resulted four sub-sequences are pair-wisely grouped to form two descriptions. Considering that two sub-sequences in a description have strong spatial correlation and temporal correlation, so the best modes and motion vectors in corresponding macro-blocks are basically the same. In this paper, only one sub-sequence per description is coded by a H.264 coder, and the other sub-sequence is coded by using the best modes and motion vectors of the aforementioned encoded subsequence. Consequently, only one sub-sequence per description needs to perform mode decision, reducing the computational complexity and bit rates. The experimental results have shown that at moderate and high rates, the proposed algorithm achieves a higher coding quality compared with other H.264 based MDC algorithms.