A new speed control strategy is presented for high performance control of a Permanent Magnet Synchronous Motor (PMSM). A self-tuning Neuro-PID controller is developed for speed control. The PID gains are tuned automatically by the neural network in an on-line way. In recent years, the researches on the control of electrical machines based on ANN are increased. ANN's, developed controller in this work, offer inherent advantages over conventional PID controller for PMSM, namely: Reduction of the effects of motor parameter variations, improvement of controller time response and improvement of drive robustness. The PMSM drive system was simulated by using MATLAB 5.0/Simulink software package. The performance of the proposed method is compared with the conventional PID methods. At the result, the control based on self-tuning Neuro-PID control has better performance than the conventional PID controller.