A fuzzy neural network (FNN) position controller is proposed to control the ultrasonic motor (USM) servo drive. The FNN controller is trained on-line using the proposed delta adaptation law. Moreover, a new driving circuit for the travelling-wave type ultrasonic motor (USM), which is a push-pull DC-DC power converter and a two-phase series-resonant inverter combination, is presented. First, the network structure, the on-line learning algorithm and the proof of convergence of the learning algorithm of the FNN are described. Next, the operating principles of the proposed driving circuit for the USM are described in detail. Then, the FNN position controller is implemented to control the USM drive to reduce the influence of parameter uncertainties and external disturbances. The effectiveness of the proposed driving circuit and FNN controller is demonstrated by some experimental results.