In view of the non-stationary features of vibration signals of gear and the difficulty to obtain a large number of fault samples in practice, a fault diagnosis scheme based on empirical mode decomposition (EMD) energy entropy and support vector machine is put forward in this paper. Firstly, original acceleration vibration signals are decomposed into a finite number of stationary intrinsic mode functions (IMFs); the energy of vibration signal will change in different frequency bands when fault occurs. Therefore, to identify the fault pattern and condition, energy feature extracted from a number of IMFs that contained the most dominant fault information could serve as input vectors of support vector machine. Practical examples show that the diagnosis approach put forward in this paper can identify gear fault patterns effectively.