Introduces an adaptive fuzzy filter based on approximate entropy and applies the filter to the digital filter of rocket telemetry signal in the wideband noise. The adaptive fuzzy filter algorithm is discussed, and the experimental filtering results of rocket telemetry signal in white noise are given. Both analysis and experiment show that the parameters of this novel filter can adjust itself according to the complexities of the signal. It filters better than the average subsection fuzzy filter for the non-steady stochastic signal in the wideband noise.