The power distribution range of a 4WD vehicle directly depends on the electromagnetic control unit of its electromagnetic multi-plate clutch. To study their relation, it is essential to build a dynamical model, since one can hardly control the electromagnetic multi-plate clutch accurately or provide valid input variables to the vehicle. In this article, we firstly build a mechanical model of the electromagnetic multi-plate clutch based on electromagnetics and system dynamics, which, together with the dynamics model (7 degree of freedom) of the vehicle, forms a model for the torque distribution mechanism in Matlab/Simulink. Next, we apply neural network to the linear 2-dof half vehicle model to optimize torque distribution control system, and study the effect of torque distribution on rotational speed difference, vehicle speed, sideslip angle and yaw rate, where the change of torque distribution is caused by the change of electric current in the electromagnetic multi-plate clutch under different pavements, vehicle speeds and steering situations. The simulation results show that the output torque of electromagnetic multi-plate clutch is in direct proportion to the electric current, and that it can control rotational speed difference in less than 0.5 s, maintain the vehicle speed, and achieve the expected value of sideslip angle and yaw rate. Finally, we compare the data obtained from a real vehicle to that of the simulation, and find them well matched, thus confirm our simulation model to be valid. It can provide theory basis for four-wheel drive transmission design of electromagnetic multi-plate clutch. © 2017, Science Press. All right reserved.