One of the control problems drawing engineers' and researchers' attention is the tracking control with predicted reference signals. The applications of this control problem are found in electric power control systems, automatic driving systems, and so on. The difficulty of the tracking control with the predicted signal is that the prediction errors lead to performance degradation of control systems. This paper aims to shape the predicted signals based on past information of reliable reference signal and system dynamics information to reduce the performance degradation caused by prediction errors. First, a signal shaping mechanism called prediction governor was proposed, and its optimal design problem was formulated. Then, an optimal solution for the design problem was analytically derived and the performance limitation of the prediction governor was clarified. The proposed optimal prediction governor is tailor-made for a given control system. By adding it to the control system, the influence of the prediction error generated by a given prediction tool can be minimised. Finally, the usefulness of the proposed optimal prediction governor was illustrated through an electric power balancing control problem with a microgrid system.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.