Abstract : Motion compensated wavelet transform (MCWT) coding is emerging as a promising framework for efficient and scalable compression of video. Moreover, the motion−compensated lifting scheme is known to be an efficient way to implement wavelet transforms for applications dedicated to video coding. In order to increase the performances of a wavelet−based video coder, we minimize the influence of some badly estimated motion vectors on the motion−compensated wavelet transform. We propose in this paper a novel and adaptive method for the implementation of the lifting scheme. More precisely, we closely adapt the lifting steps to the motion by sampling the mother scaling function of a multiresolution basis according to the energy of the motion. Then, we compute the new lifting operators. First experiments of this approach give promising results.