Gears are the most essential parts in rotating machinery,its health status is a significant index to indicate whether machines can run continually or not.So,an automatical and accuracy way for gear early stage fault diagnosis is very important in engineering practice.A novel gear fault diagnosis method based on Euclidean distance technique（EDT）is proposed.The energies of frequency bands are used as the fault feature paramdter in this method.Then,the euclidean distance between the test sample and each health status are calculated through the matrix which consist of the energy values.The class which is the test sample belongs is decided by the minimum distance value.The effectiveness of this methodology is demonstrated using an experiment conducted on a real gearbox.The results show that EDT can diagnose the gear crack fault with a higher accuracy rate than other similar method.As the same time,band energy is a sensitive feature parameter in gear crack fault diagnosis and can be used later.