As we have seen that Huffman coding has been widely used in data, image, and video compression. In this paper novel maximal prefix coding is introduced. Relationship between the Huffman coding and the optimal maximal prefix coding are discussed. We show that all Huffman coding schemes are optimal maximal prefix coding schemes and that conversely the optimal maximal prefix coding schemes need not to be the Huffman coding schemes. Moreover, it is proven that, for any maximal prefix code C, there exists an information source I = (∑, P)such that C is exactly a Huffman code for I. Therefore, it is essential to show that the class of Huffman codes is coincident with one of maximal prefix codes. A case study of data compression is also given. Comparing the Huffman coding, the maximal prefix coding is used for not only statistical modeling but also dictionary methods. And it is good at applying to a large information retrieval system and improving its security.