This paper describes a neural network controller for permanent magnet brushless dc (PMBLDC) motor drives. The objective is to control the rotor angular speed to follow an arbitrarily selected trajectory. The main feature of the proposed controller is neural network, which captures the unknown inverse dynamics of the motor. Performance of the neural network controller is evaluated by simulating it in a typical brushless dc motor model. Finally neural network performance is compared with the conventional controller performance to show that NN performance is good.