Proposed in this paper is a noise suppression algorithm for chaotic mapping on the basis of nonlocal mean． According to the characteristics of chaotic mappings，this algorithm obtains the optimal filtering parameters (including patch size，search neighborhood and bandwidth) of the nonlocal mean applied to the noise suppression of chaotic mappings via experimental analysis． Simulated results show that the proposed algorithm outperforms such existing methods as phase space estimating projection，extended Kalman filtering and unscented Kalman filtering in terms of Gaussian noise suppression; and that it effectively suppresses chaotic mappings at different noise levels．