Abstract: With the development of Internet, frequent pattern mining generalizes to more complex patterns like tree mining and graph mining Such applications arise in complex domains like Bioinformatics, web mining, etc In this paper a novel algorithm, named ESPM （expanded subtree pattern miner）, is presented to discover frequent subtrees from ordered labeled trees Unlike previous works, the work of distinguishing isomorphism is left in the later part of the algorithm, which minimizes the cost of the whole process The performance of the algorithm is evaluated with experiments on synthetic and real datasets The experimental result shows that the algorithm can do the job well and is better than previous algorithms Finally the potential improvement of ESPM is mentioned