Abstract: Periodic impact signal is the key index to diagnose rolling bearing defects. It is very important to extract the impulse of defects effectively for accurate detection of bearing faults. However, under the influence of noise, the impact characteristic of rolling bearing fault signal is always submerged. In order to extract periodic impact signals from low SNR signals, this paper used a class of third-order Teager energy operators to diagnose rolling bearing faults. The paper used the third-order Teager energy operator to calculate the instantaneous total energy for the fault vibration signal, performed the spectrum analysis with Fourier transform. The fault features could be extracted from the thirdorder Teager energy spectrum.The results show that the third-order Teager energy spectrum can clearly show the fault characteristics, and the effect is obviously better than the traditional spectrum analysis method, envelope spectrum analysis method and second-order Teager energy spectrum method.