In order to measure the difference of models，a method of computing three dimension model’ssimilarity based on self-adaptive global best harmony search algorithm is proposed. Two models’face similarity matrix is constructed according to the number of face’s edges and adjacency relationship of faces. Self-adaptive global best harmony search algorithm is used to find an optimal face matching sequence between source model and target one from face similarity matrix. Based on an optimal face matching sequence，similarities between source model faces and target ones are summed to compute two models’similarity. Experimental results show that the proposed method can measure two models’difference more accurately.