Abstract: The further development of expert systems needs adopting other advanced techniques.Neural networks and fuzzy systems are completely new ways which help to overcome the bottlenecks of conventional expert systems,such as the difficulties in acquiring knowledge and the uncertainties of the partial information.The application of fuzzy neural networks in expert systems is introduced.The principle,method,structure and realized techniques of the expert system based on fuzzy multilayer perceptron （FMLP） are described.An expert system based on FMLP is applied in classifying remote sensing images,thus,achieving satisfying result.