Abstract: A new diagnostic method has been developed,which applies the fuzzy self-organizing neural network to aeroengine fault diagnosis.In this method,the relationships between the fault causation and fault symptoms are established aecording to the fuzzy mathematics.Then the mapped 2-dimension pictures of the sample fault set and actual fault are obtained by a self-organizing neural network.At last,the reasonable conclusions are drown from the comparison between them.The case study of an engine has verified the practicability of this method.The proposed method features a simple structural algorithm,supervision free self-study and lateral association,etc.