Based on the Lyapunov stability theorem, an optimal model reference adaptive control (OMRAC) scheme designed for a class of multi-input (MI) systems with input non-linearity for solving robust tracking problems is presented in this paper. The proposed control scheme contains an optimal controller designed for nominal systems, and an adaptive mechanism used to automatically adapt the unknown upper bound of perturbation. The asymptotical stability of tracking error is guaranteed for the controlled system. A numerical example is given for demonstrating the feasibility of the proposed control scheme.