Teager energy operator has been widely used in the detection of bearing faults. It tracks and analyzes the instantaneous energy and has a unique advantage in detecting impulses in a signal. However, its applications to sound signals of the train wheel bearings have not seen many reports. As such, a scheme of fault diagnosis detection of rolling bearing using demodulation is introduced based on the Teager energy operator. First of all, the concept and demodulation principle of Teager energy operator were presented. And then the sound signals collected from 197726TN wheel bearings of running trains were analyzed. The spectrum was thus obtained from the resulted instantaneous energy waveform, which envelop was analyzed and obtained, and the bearing health condition could be readily determined. The comparison with Hilbert transform envelope demodulation was highlighted that Teager energy operator demodulation could reveal the characteristics of fault signal and determine the fault type much effectively. The effectiveness and merit of the method is verified by the analysis of field data of sound signals in wheel bearings.