In this paper, aiming at problem of synchronization for perturbed chaotic cellular neural networks with time-varying delay, introduce a self-adaptive synchronization method. Using the Lyapunov stability theory, the Barbalat lemma and self-adaptive synchronization control method, we get an easy realized adaptive synchronization control law to realize the global asymptotical stability of the error system. The control law has the better robustness. And the method is used to secure communication by chaos masking. The simulation result verifies the effectiveness of the synchronization method.