The two-phase hybrid stepping motor is a nonlinear, multi-variable and strongly coupled system. Aiming at the problems of simple open-loop control structure and low positioning accuracy of two-phase hybrid stepping motor, a backstepping control method based on radial basis function (RBF) neural network is proposed. The control method overcomes the shortcomings of traditional backstepping control for nonlinear control, uses the universal approximation property of RBF neural network to compensate for the uncertainties in motor operation. Meanwhile, Gaussian basis function and adaptive law are introduced, which can make the RBF neural network better compensate for the uncertainties in motor operation. The combination of neural network and backstepping control methods effectively improves the position tracking accuracy and steady-state performance of the two-phase hybrid stepping motor control.