In this paper, an application of genetic algorithm for on-line tuning of a feedforward controller in nonlinear servo systems is described. The parameters in the feedforward controller are regarded as genes and tuned by a genetic algorithm so that the tracking error converges to zero. To achieve high speed tuning, new genetic operations such as a few individuals, quick estimation, instant selection and intentional mutation are introduced. After the tuning process is completed, the nonlinear characteristic of the plant is linearized and an equivalent dynamic inverse is made in the feedforward path. The experimental results demonstrate the effectiveness of the proposed feedforward controller for improvement of tracking performance of the nonlinear speed servo systems.