AbstractMud pumping is a major problem for high-speed rail (HSR) slab tracks because it can cause deterioration of track alignment and further compromise operation safety and ride comfort of passengers. Timely inspection for mud pumping is crucial to the operational safety of high-speed rail networks. This paper proposes a novel detection method for mud pumping defects based on the measurement data of track geometry. The time and frequency features of the surface irregularity signal corresponding to the mud pumping sites were first analyzed. The concave anomalies reflecting the features of mud pumping and the peaks at the wavelength of 5 m and its octave on the power spectrum density (PSD) curve were revealed. On this basis, to automatically detect mud pumping defects, a multiscale signal decomposition method was employed and defect-sensitive features were extracted. A mud pumping index (MPI) indicating the severity of mud pumping was then established, and a three-level management scheme for mud pumping maintenance was developed accordingly. The performance of the proposed detection method was verified against visual inspection recorded on an in-service HSR line, which was constructed with the China Railway Track System (CRTS) I slab track. The results show that the proposed detection method can effectively locate and assess the mud pumping defects with acceptable accuracy for HSR track maintenance needs.