Abstact In order to implement the precise position control of linear ultrasonic motor, an adaptive control scheme based on radial basis function neural network for is proposed. Due to the mechanism of the linear ultrasonic motors, their operation status will be subject to friction, strong nonlinear and time-varying uncertainties interference. In order to approximate these uncertainties effectively, a radial basis function neural network is adopted. In order to improve the adaptive preference, firstly, the number of the hidden units and parameters of the network is obtained using orthogonal least square algorithm with the training data collected from experiments, then, the weights between the hidden layer and the output layer are updated according to the recursive least square algorithm. The experimental results show the proposed controller is better than PID and error back propagation neural network controllers, and has a good anti-interference capability.