In this paper, the synchronization problem for chaotic neural networks with sampled-data control and actuator saturation is investigated. By constructing a suitable time-dependent function and utilizing a modified free-matrix-based integral inequality, a sampled-data synchronization criterion for chaotic neural networks is derived as the framework of linear matrix inequalities. Based on the first result, a design method of sampled-data controller subject to actuator saturation for chaotic neural networks is introduced through an optimization method which enlarges the set of admissible initial conditions. The superiority and validity of the proposed results will be verified through comparing with the existing results in a numerical example.