This paper presents new bottle-neck diagnosis method in a regional economic system based on fuzzy adaptive inference. This method can partly solve the problems such as many obstacles, complicated influence factors, and fuzzy characteristics in the regional economic development. It combines the fuzzy logic with neural network and has all the advantages which they both have. It can realize bottle-neck diagnosis in a short time. It has obvious advantage in terms of knowledge acquisition, parallel inference, adaptive learning, memory recalling and generalization of knowledge. Simulation results show that the application of fuzzy neural network inference in the regional economic bottle-neck diagnosis system is in accordance with the local authority's usual practice. Thus, this method is more valid than the method based on rules.