Similarity metric of process model is an indispensable task in business process management, which is widely used in many scenarios such as organizations merging, user requirements change, and model repository management. This paper focuses on the similarity metric algorithm based on the principal transition sequences (PTS) and puts forward improvement scheme by defining the complete firing sequences to express model behavior, matching fining sequences by A^* algorithm with pruning strategy and proposing new similarity metric method. Experiments show that this method improves rationality than existing behavior-based similarity algorithms.