A collision severity minimization strategy based on hierarchical structure of decision layer and motion control layer was proposed aiming at how to effectively reduce vehicle collision severity and ensure high real-time response in unavoidable collision scenarios. The post-collision instability risk was estimated by using the four-degree-of-freedom collision model and the vehicle state before the collision by considering that the offline trajectory library was generated by solving the optimal control problem under the constraints of vehicle dynamics. The collision severity assessment model was used to evaluate the collision risk and post-collision instability risk of the trajectory library. The optimal trajectory was determined from the trajectory library in a very short time during decision making. A collaborative controller for trajectory tracking and yaw stability was established based on model predictive control in the motion control layer in order to maintain trajectory tracking accuracy and vehicle stability. The effectiveness of the proposed collision severity minimization strategy was verified by simulation in different scenarios. The simulation results show that the proposed collision severity minimization strategy can effectively reduce vehicle collision severity while ensuring vehicle stability under different working conditions.