Abstract: In response to the non-stationary characteristics of the network delay, a new adaptive playout algorithm based on wavelet transform is proposed. This algorithm uses wavelet transform to decompose the non-stationary delay series into several stationary components, and then all these components are predicted by,auto regressive(AR) models with different adaptation rates. Subsequently, the final end-to-end delay is obtained by wavelet reconstruction with the predicted components. Simulation results showed that the prediction precision of the new algorithm is about 5~14dB higher than that of the conventional AR algorithm, and about 1.5~5dB higher than that of the latest differential auto regressive (DIAR) algorithm.