This paper proposes an intelligent Cerebella Model Articulation Controller (CMAC) with the sliding mode control. Generally speaking, the CMAC is first trained from an existed controller; then the CMAC can be used to instead of the original controller for achieving a fast operating system performance. Moreover, it can be easily realized through FPGA and operated in the requirement of real-time control cases. However, the overall performance of control system depends on the trained CMAC; usually, it is not easy to yield the same performance as the original controller. Therefore, to improve the performance without learning process maybe is a good idea Finally, the DC motor speed control is employed to illustrate the performance and applicability of the proposed control scheme.