A method to simplify fuzzy logic system (FLS) structure was presented based on rule sensitivity. With introducing the concept of rule sensitivity, a new function was constructed and added into objective function to get a penalty function.In this way, a new performance index was obtained. Then, fuzzy logic system parameters were modified by optimizing the performance index. On the above basis, as many as possible sensitivities can be approximated to zero, while keeping outputs of the fuzzy logic system approximate to the desire ones. Finally, the fuzzy logic system was simplified by eliminating those rules with approximately zero sensitivities. The efficiency of the method was verified by simulation results.