It is important and valuable to reveal the spatial pattern, changing characteristics of nitrogen and phosphorus emissions from agricultural sources and the relationship between them and water environment for rational allocation of investment in agricultural non-point source pollution control and optimization of regional agricultural structure. Based on this understanding, a series of parametric models were established to estimate the agricultural source nitrogen and phosphorus emissions in China from 1990 to 2017. Then, the changing characteristics of nitrogen and phosphorus emissions from agricultural sources were studied from three aspects including total emissions, nitrogen and phosphorus emissions per unit area and migration of emission center, and the relationship between nitrogen and phosphorus emissions from agricultural sources and water environment was revealed through correlation and regression analysis. The study found that the nitrogen and phosphorus emissions from agricultural sources in China were distributed randomly between provinces, indicating that there was no significant interprovincial dependence of nitrogen and phosphorus emissions, which were mainly influenced by internal factors of each province. Throughout the study period, the provinces with the highest annual total nitrogen and phosphorus emissions were Henan, Shandong, Sichuan, Hubei, Hebei, Anhui and Jiangsu provinces. The pattern was relatively stable, but the emission center experienced a reciprocating transfer from east to west and had an overall transfer trend in the northwest direction. Among them, the center of gravity of nitrogen and phosphorus emissions shifted 33.94 km and 52.31 km respectively in the direction of north by west 33.52° and 15.45°. Based on the pattern research and combined with pattern characteristics of spatial dimension and changing characteristics of time dimension, nitrogen and phosphorus emissions of 31 provincial-level regions were divided into 8 categories, namely, high emission growth type, high emission stable type, high emission inverted "U" shape type, medium emission growth type, medium emission inverted "U" shape type, low emission growth type, low emission inverted "U" shape type, and low emission approximate "U" shape type. Furthermore, the water quality data from automatic monitoring stations in 2017 were used to verify the relationship between nitrogen and phosphorus emissions and water environment. The results can be used as the basis for identifying key regions in agricultural non-point source pollution control.