Abstract: Nitrogen pollution in groundwater systems in Huai River Basin has drawn a lot of attention. In order to trace the source of nitrogen pollution in groundwater, 2002-2017 agricultural statistics data for 35 cities in the Huai River Basin agro-ecosystem were collected. A nitrogen balance model was set up based on nitrogen input and output in Huai River Basin, and it was used to calculate nitrogen surplus and intensity in the basin. Nitrogen input included input from fertilizers, humans & animal excreta, atmospheric deposition, biological fixation, seed nitrogen and straw nitrogen. Nitrogen output included crop harvest, denitrification and ammonia volatilization output. Also, combined nitrogen surplus and leaching coefficient, the nitrogen leaching model was built to quantitatively estimate the amount of nitrogen leaching into groundwater bodies from agro-ecosystem in Huai River Basin. The results showed that average nitrogen input in Huai River Basin agro-ecosystem was up to 10 050 100 t·a^-1 for the 2002-2017, fertilizer input was the largest source of this amount nitrogen input and it accounted for 52.76%. Average nitrogen output was up to 7 064 300 t·a^-1 for the period 2002-2017, crop harvest was the largest amount output of this amount nitrogen and it accounted for 87.29%. Nitrogen surplus and nitrogen source intensity decreased gradually with time for the period from 2002 to 2017. Nitrogen source intensity result was the same with previous studies, which ensured the accuracy of the results. At the regional aspects, the city in Henan Province had the highest nitrogen source intensity, while the cities in Shandong and Anhui Provinces had the lowest nitrogen source intensity. The amount of nitrogen that leached into the groundwater in Huai River Basin agro-ecosystem was 2.622×10⁵-4.171×10⁵ t·a^-1, with the highest amount in 2006. The average nitrogen amount in groundwater was 3.141×10⁵ t·a^-1 for the period from 2002 to 2017, which caused a large pollution load in the water environment. F statistic and ρ value tests in SPSS 21.0 gave the relationship between the actual nitrate concentration in groundwater and the estimation nitrogen amount leaching into the groundwater. Finally, the estimated and observed values passed significance test, with a correlation coefficient of 0.517, which confirmed the accuracy of the model. Nitrogen input as chemical fertilizer input and nitrogen output as crop harvest were respectively the main input and output factors in the study area. The most serious pollution was in 2006. The study provided important data needed to solve non-point agricultural pollution with important practical implications for the prevention and control of nitrogen pollution in groundwater.