The large distributed interactive simulation system which based on the publish/subscribe model has considerable data, it needs to disseminate accurate data to the interested users quickly, so improving the match efficient is a very important method to solve this problem. Based on the study and comparison of current matching algorithm, we have proposed a predicate priority based event matching algorithm(hereinafter referred to as PPEM). Performance analysis and detailed experimentations are carried out to verify the effectiveness of PPEM. When the number of subscriptions is 10000, the average time of matching one event is only 22.6ms and when we change the type of data distribution or increase the number of event attribute, the average matching time of PPEM is almost unchangeable comparing with bucket and multi-dimension algorithms. All the results show that PPEM achieves high matching efficiency and also has good scalability and robustness.