The fault diagnosis method based on waveletneural network in analog circuit was presented. The problems on feature extraction in analog circuit, data pre-processing, training and testing of network, fault pattern classification were all discussed here. Fault features were extracted by circuit simulation, after data preprocessed by wavelet-neural network, which can be constructed into samples aggregation, the samples for training and testing were used to train and test neural network respectively, the uniform pattern samples were used to diagnose faults, the fault diagnosis in analogy circuit was realized effectively. An illustration validated this method.