In order to solve the problem that the traditional variable universe fuzzy controller for semi-active suspension dependents on empirical rules, an improved variable universe T-S fuzzy control strategy based on fuzzy neural network is proposed. Firstly, according to the testing data of the magneto-rheological (MR) damping force characteristic, the damping force of MR shock absorber based on adaptive fuzzy neural network and the half vehicle MR semi-active suspension system is established. Secondly, the T-S fuzzy controller for suspension system is developed, meanwhile in order to adjust the variables of T-S fuzzy controller in real time, the fuzzy neural network structure is used to describe the change of the scaling factor. Finally, the effectiveness and feasibility of the proposed control method are verified by the simulation results.