Agricultural nitrogen (N) and phosphorus (P) emissions to water bodies remain largely unknown in China, mainly due to the lack of reliable data sources and quantification tools. In this study, we constructed a grid-based NUFER (NUtrient Flow in food chains, Environment and Resources use) model in order to quantify a high-resolution agricultural N and P emissions to water bodies in Haihe Basin in 2012, based on data collected from county-level statistics, farm interview, and spatial data of topography, climate, soil texture, and land use. We also explored the mitigation strategies in 2030 via scenario analysis. The results showed that total agricultural N emission to water bodies in Haihe Basin was 1079 Gg N in 2012, of which cropland contributed 54%; total agricultural P emission to water bodies was 208 Gg P, livestock contributed 78%. There were large spatial variations in agricultural N and P emissions. Overall, the plain areas accounted for around 80% of the total agricultural N and P emissions to water in 2012. The highest N and P emission intensities were 10 t N km−2 and 2 t P km−2, respectively. N and P emissions were significantly related to anthropogenic factors (such as the livestock density and cropland) in the plain areas; whereas in mountainous areas, both anthropogenic and natural factors (e.g., slope deviation and soil texture) significantly affected N and P emissions. Our scenario analysis suggests that agricultural N and P emissions can be reduced by up to 45% and 77%, respectively for N and P in 2030, via improved agricultural and environmental policies, technologies and managements. The prohibition of direct animal manure discharge to the water system seems to be the most effective measure to mitigate the emissions. Our study provided a high-resolution agricultural N and P emissions to the water bodies of Haihe Basin and identified the most effective options to reduce these emissions in highly intensified agricultural areas.