Association rule mining algorithms can be used to discover all item associations (or rules) in a dataset that satisfy user-specified constraints, e.g., minimum supports (minsup) and minimum confidence (minconf). Since only one minsup is used for the whole database, it is implicitly assumed in the model that all items are of the same nature and/ or have similar frequencies in the data. This rarely applies in reality, so, based on an FP-Tree, a new algorithm is proposed called multiple minimum supports for discovering maximum frequent item sets algorithm (MSDMFIA). The algorithm allows users to specify multiple minsups to reflect item natures and various frequencies, which resolves bottlenecks in traditional algorithms, e.g., the frequent generation of candidate itemsets and database scanning. Experimental results show that functionality and performance of the proposed algorithm is significantly improved compared with most others.