In this paper, variable precision rough set theory is introduced to logistics center location problem. Firstly, an index system is built for logistics center location. Secondly, a discretization method for continuous decision table is put forward. In this method, some crisp ldquocutsrdquo are got based on information entropy, and then to be carried out fuzzy operation. In both crisp discrete intervals and fuzzy discrete intervals, the crisp ldquocut intervalsrdquo are more important than the single ldquocutsrdquo. Thirdly, information entropy is introduced to the attributes reduction process of rough set. And then the improved attributes reduction way is used to elaborate the knowledge discovery of logistics center location from the index system. Finally, the rules mined from decision table can be used to instruct the decision support system for logistics center location.