Satellites constantly face the turbulences and external torques and so they may exhibit chaotic behavior and what is expected will not happen. On the other hand, it is so difficult or impossible to do some missions by only one satellite. These missions include the Earth observation for different purposes, observation of objects in space, etc. that can become more accurate by several synchronized satellites and also the costs of setting up and control with establishment of small satellites will be much cheaper. In this article, neural network controllers are used for synchronization of two chaotic satellites (which called master and slave satellites). Neural network controllers such as NARMA-L2 and Predictive controllers have been applied very successfully in the identification and control of dynamic systems. Both of the neural controllers have very good performance in the synchronization of two chaotic satellites, but each one has its advantages and disadvantages which in this article they will be described. Finally, results of the simulation have shown the very good performance of these neural controllers in the ideal synchronization of the salve chaotic satellite with the master chaotic satellite.