Abstract: A new, decentralized PI (proportional integral) controller design method is propsed for high-dimensional multivariable processes. By considering the steady and dynamic information from multivariable systems, an analytical expression of the parameters of the PI controller is derived by employing a closed-loop transfer function matrix under unit feedback. The proposed decentralized controller reflects the relationship between its parameters and the steady and dynamic information of the original processes. Compared with existing methods, it can improve the dynamic performance of the control system. Even in the case of model mismatch, the control system can maintain better control performance and is more robust. Simulation experiments have been made on several typical controlled objects, which demonstrate the practicality, effectiveness, and superiority of the proposed controller.