As the core component of vehicle intelligent suspension system, the accuracy of MR (magnetorheological) damper’s inverse model is one of the main factors affecting the vibration attenuation performance. In order to improve the modeling accuracy of MR damper caused by hysteretic nonlinearity for the all-terrain vehicle suspension system, a fuzzy T-S modeling method based on improved Gath-Geva clustering is proposed in this paper. First of all, the output force generated by MR damper under various excitation frequency, amplitude and current is tested and analyzed by MTS machine. Then, based on the test results and T-S fuzzy reasoning method, the inverse model of the damper is established, while the improved Gath-Geva clustering method is employed to identify the parameters of the T-S fuzzy model. The relationship among excitation displacement, velocity, damping force and control current is obtained, and the modeling accuracy and parameter identification speed are improved. Finally, the proposed fuzzy nonlinear modeling method is verified by the non-modeling data obtained in the experiment. The results show that the proposed method has high prediction accuracy for the control current of the damper, and the root mean square error between the predicted current and the experimental value is only 0.008 8 A.