Abstract: In this paper, for the incomplete information system, the definition of the missing rate of attribute value on granularity is given, and the combination importance of granularity is defined by combining the importance of granularity with the missing rate of attribute value on granularity. Then, a new algorithm of granulation reduction is designed with the combination importance of granularity as the heuristic function, the reduction results of the low missing rate of attribute value can be obtained while ensuring the reduction rate by adjusting the parameters. Finally, the experimental results verify the effectiveness of the proposed algorithms in this paper.