As a concept of development, the sustainable and coordinated development of agricultural system has become a consensus, which has been attached great importance by our government. Using the panel data of agricultural production in China from 2004 to 2015, this paper employs the entropy method and exploratory spatial data analysis to analyze the coordination degree of sustainable development of China's agricultural system. The results showed that (1) from 2004 to 2015, the coordination degree between the sustainability of agricultural subsystems in various provinces (municipalities and autonomous regions) in China presented an unstable and dynamic trend, and its coordination value was low. (2) From the perspective of coordination degree distribution, the provinces (municipalities and autonomous regions) whose coordination degree was "comparatively coordinated" and "coordinated" were mainly distributed in the eastern and southeastern regions of China, while the coordination degrees among the agricultural subsystems of most other provinces (municipalities and autonomous regions) were "uncoordinated". (3) From the perspective of spatial autocorrelation, the coordination degree between the sustainability of agricultural subsystems in each province (municipalities and autonomous region) had a positive whole autocorrelation and the degree of correlation gradually increase. In most of the provinces (municipalities and autonomous regions), the local autocorrelation of the degree of coordination between system sustainability was not significant. The spatial polarization of the degree of coordination between the sustainability of agricultural subsystems was not obvious being at a lower level of equilibrium. This paper analyzes the coordination degree between the sustainability of agricultural subsystems from the perspective of spatio-temporal dimension and spatial correlation, and provides policy recommendations for improving the sustainable development level of the Chinese agricultural system.