Abstract: In this paper, the towns of Fu’an are taken as spatial units and the gross agricultural production (GAP) per capita of each unit is chosen as the single evaluation index. Based on the theory and method of spatial autocorrelation analysis, the author draws the overall and local autocorrelation graphs of Moran’s I exponent to study the spatial autocorrelation characteristics of agricultural economic pattern using GeoDa, DPS and ArcGIS software. The results showed that the difference of town agricultural economic structure in Fu’an has mainly appeared in southeast- northwest direction since 2000. There are obvious polarization and aggregation trends in the east- west direction. The developed agricultural economic towns are mainly concentrated in the southeast of Fu’an including Wanwu, Xiabaishi and Xiwei. In addition, the backward agricultural economic towns are Banzhong, Xitan, Muyang, and Chengguan. It is predicted that the town agricultural economic pattern of Fuan will not change significantly in 2020, and agricultural economic difference will be narrowed.