This paper uses the super-efficiency SBM model of unexpected output to measure the agricultural productivity of China's provinces from 2000 to 2020,and then uses social network analysis and quadratic assignment procedure(QAP)method to identify the spatial correlation characteristics and influencing factors of agricultural productivity.The results are as follows:The spatial correlation of provincial agricultural productivity in China is getting closer,and the stability of network structure is strong.Henan,Shandong,Hubei and other large agricultural production provinces are the central actors in the spatial correlation network,while Hainan,Shanghai,Tianjin and other provinces with more geographical locations or small agricultural production scale are the marginal actors.Narrowing the differences in agricultural economic development level,agricultural technology level,the level of agricultural human capital and the level of marketization can help strengthen the spatial correlation of agricultural productivity among regions,and it is easier to establish spatial correlation between geographically close provinces.