Abstract: It is significant for automobile security to study how to monitor operating state of automobile gearbox and detect incipient faults as soon as possible. Characteristics analysis for gearbox vibration signals captured from vibrating acceleration sensors based on empirical mode decomposition (EMD) and Teager energy spectrum was proposed. The fault detection signal was firstly decomposed into intrinsic mode functions (IMF) with the EMD method. Then, Teager energy spectrum of each intrinsic mode function was obtained using Teager energy operator. Finally, the time-frequency fault characteristics of a gearbox were analyzed by means of the coefficient-energy spectrum values of intrinsic mode functions. Experiment result has showed the feasibility and efficiency of the EMD and Teager energy characteristic method in fault location and fault information exstraction, and that the algorithm is very reliable to be implemented for fault detection.