In this paper, a H-infinity synchronization controller is studied based on neural network disturbance observer for two uncertain chaotic systems. The disturbance observer is designed using the radial basis function (RBF) neural networks, and the parameter's update law is presented for monitoring the compound disturbance well. Based on the output of the disturbance observer, the H-infinity synchronization controller is proposed which can make the synchronization error convergent to zero under appropriate parameter adaptive laws. Finally, a simulation example is given to validate the availability of the proposed synchronization control method.