Understanding the spatiotemporal distribution of nonpoint-source pollutant river loads from catchments and their influences on the water quality along river networks in a large river basin is important for land management and water quality protection. In this study, a geomorphology-based nonpoint-source pollution (GBNP) model was applied to the upper Yangtze River basin (YRB) to analyze the nitrogen loads into the river network, as well as its influences on river water quality. Based on the simulation, we analyzed the spatial and temporal distribution of nonpoint-source nitrogen loads into river networks and the nitrogen retention along the river network. The results illustrated that the annual total nitrogen (TN) loads from the fields/hillslopes into the rivers ranged from 55 × 104 t to 392 × 104 t on average during 1990–2012 in the upper YRB. The export coefficient (the ratio of river TN load to TN input of the same sub-catchment) over the upper YRB varied from 0.04 to 0.82 and had a mean value of 0.26. The spatial variability in the annual TN load (TN load intensity) significantly influenced the export coefficient magnitudes in different tributaries. The river nitrogen retention ratio varied from 3.84% to 85.50%, and Minjiang (MJ) had the highest value among all tributaries. The nitrogen retention ratio was higher in spring and winter than in summer and autumn in all tributaries. Both temperature and stream flow conditions (discharge and velocity) were the major factors that influenced river nitrogen retention. In general, a higher TN load from the catchment resulted in a higher river TN concentration due to the “lift-up and carry-away” functions of the catchment rainfall-runoff processes. However, when the net anthropogenic nitrogen input (NANI) was low, the TN load was also relatively low, and the high flow could dilute the river nitrogen concentration. When the TN load was adequately high, the river flow dilution could disappear, and the river TN concentration had the same pattern as the river discharge.