Abstract: Magnetic flux leakage (MFL) testing signals obtained from oil pipelines are usually contaminated by various sources of noises, and therefore the detectability of defect in MFL data is considerably reduced. The fuzzy lifting wavelet packet was presented and used to solve the problem of noise reduction from the magnetic flux leakage (MFL) inspecting signals. Under the condition of three kind of threshold processing mode: soft-threshold，hard-threshold and fussy-threshold, the de-noising performance of proposed method was compared with traditional wavelet packet. The simulation results showed that the fuzzy lifting wavelet packet de-noising method had the characteristic of fast operational speed and good de-noising effectiveness, so it was a valid on-line processing method of MFL inspecting signal.