According to fuzziness and complexity of mechanism wear fault, a kind of intelligent diagnosis system for mechanism wear fault based on fuzzy-neural network is put forward. The structure and learning method of fuzzy-neural network are introduced. This paper analysises how to create the characteristic vector of wear particles and standard wear particles spectrum by combine the characteristics of mechanism wear fault, and the fuzzy-neural network of mechanism wear fault intelligent diagnosis is built. Mechanism wear faults can be diagnosed to apply the fuzzy-neural network and fault causes are determined. This system is developed by applying MATLAB software and the GUI functione of MATLAB for the simplicity and intuition of diagnosis.