Abstract: Association rule mining which is used to find the correlation of items is an important research direction in data mining.FP-growth algorithm greatly reduces the search time without generating candidate itemsets by constructing FP-tree to find frequent itemsets.The drawback of it is taking a lot of memory space.Based on the thinking of FP-growth algorithm，an algorithm for mining frequent patterns based on FS-tree（1 item frequent sub-tree） is proposed.The algorithm reduces the space complexity significantly by splitting FP-tree into some FS-trees.The experiments show that the algorithm is effective.