Abstract: Abstract: It has been widely proved that nitrogen is vital to maintain terrestrial ecosystem balance and world food safety at global scale. However, with the increasing development of social economy and population growth, excessive synthetic nitrogen has been discharged into water bodies in the world, which is recognized as one of the most important causes of water environment deterioration and eutrophication. Within the numerous nitrogen sources, diffuse pollution from agricultural activities has been identified as the most important contributor to nitrogen loss in more and more areas. Therefore, it is more crucial and meaningful to assess precisely nitrogen loss potential and to identify its impact factors for the effectively integrated watershed management. In this study, a modified nitrogen export coefficient model was developed considering the nitrogen production, surface runoff generation, leaching potential of soil moisture, and landscape interception in the given watershed. To examine the performance of the modified total nitrogen export coefficient model, this modified result was validated by the monitoring data of water quality at the outlet of Chao River watershed, which is one of the 2 major tributaries into Miyun Reservoir in the northeast of Beijing. Then, the critical source areas (CSAs) of nitrogen loss were identified, and the impact factors of nitrogen loss were determined by the boosted regression trees algorithm. The major results were illustrated as follows: 1) For monitoring loading of total nitrogen at outlet of Chao River, the modified export coefficient model was characterized by a higher accuracy with a relative error of 8.23% compared with the traditional export coefficient model with a relative error of 18.94%. 2) The CSAs of total nitrogen loss were mainly distributed in the western areas of upper and middle stream, as well as the downstream riparian region in this watershed, where existed higher nitrogen load for intensive agricultural economic activities. Regarding to administrative scale, Fengning County (Dage and Heishanzui Town), Luanping County (Hushiha and Bakeshiying Town), Miyun District (Gaoling, Gubeikou and Taishitun Town) were identified as the CSAs of total nitrogen loss for their higher population and livestock density, and the relative lower capability of landscape interception. In particular, the high potential of total nitrogen loss was also identified in the south area and riparian zone in the Chao River watershed. 3) Based on the boosted regression trees model, by extracting 5 000 samples in ArcGIS 10.1 platform, the anthropological factors were determined as the most important impacting factors for total nitrogen loss. The contributions of the amount of total nitrogen fertilizer application, the livestock breeding scale and the topography slope to total nitrogen loss in the watershed were 54.74%, 17.48% and 16.53%, respectively. Therefore, control on the unreasonable scale of total nitrogen and livestock was necessary and recommended to reduce the severe nitrogen pollution and to protect the drinking water safety of Beijing City. In summary, the proposed total nitrogen export coefficient model can be applied to provide available information for the environmentalists, farmers and watershed managers to prevent and alleviate diffuse total nitrogen loss especially in the area lacking effective data.