Abstract: Document binary image is different from text file and general image file. It’s very difficult to reach high compress ratios using general compressing algorithms. In this paper, the characteristic of document is analyzed and a lossless mixed coding algorithm is proposed, which uses the block coding and run length coding （RLC）. Experiments prove it can increase higher compress ratios by 0.5 to 1.2 compared with dictionary based coding （ARJ, LZW, PKZIP etc.） and by 0.3 to 0.8 compared with general RLC.