A novel genetic-based neural network control for chaos is presented. The method proposed has been successfully applied to control two simulated chaotic systems by incorporating the techniques of small perturbations and the periodic control in this paper. The simulation results showed that the neural network trained by genetic algorithm can learn to produce a series of small perturbations to conver t chaotic oscillations of a chaotic system into desired regular ones. The scheme requires no knowledge about the mathematical model. Moreover, the approach is r easonably robust to noise, it can be extended to control other chaotic systems.