In this paper, we propose a new framework for the representation of deforming meshes by only updating necessary changes of the connectivity. The deforming meshes, which is also known as time-varying surfaces, are often constructed with static connectivity. To progressively represent the deforming meshes with level-of-details, people can simplify the meshes independently to obtain good simplification meshes but total different connectivity for each frame. On the other hand, people can also simplify the meshes while keeping their static connectivity, but this constraint makes the simplified meshes distorted. Hence, we present a feature-adaptive simplification scheme, which takes all timedependent information into consideration. Through the application of one-dimensional Haar wavelet, we successfully quantize the costs of all edges of the deforming meshes. Then, we apply a global optimization to minimize the meshes’ distortion while maximizing the temporal coherence. The optimization is hard in terms of computational complexity, and we therefore exploit the advantages of genetic algorithm together with dynamic programming to solve this problem. The result makes the progressive deforming meshes represented without updating the connectivity rapidly and have good simplification meshes for each frame.