A new approach to fault diagnosis of gear crack based on Ensemble Empir ical Mode Decomposition (EEMD) and Teager Huang transform (THT) is presented. First, the original times series data is decomposed into intrinsic oscillation modes by using EEMD approach. Then the Teager Kaiser Energy Operator (TKEO) techni que is applied to each intrinsic mode function (IMF). The TKEO can track the modulation energy and estimate the instantaneous amplitude and instantaneous fre quency, and the time frequency distribution can be obtained. Therefore, thech aracteristics of the gear faults can be recognized according to the time freq u ency spectrum. The experimental results show that Teager Huang transform based on EEMD can effectively diagnose the faults of the gear crack.