High external nitrogen (N) inputs can maximize maize yield but can cause a subsequent reduction in N use efficiency (NUE). Thus, it is necessary to identify the minimum effective N fertilizer input that does not affect maize grain yield (GY) and to investigate the photosynthetic and root system consequences of this optimal dose. We conducted a 4-year field experiment from 2014 to 2017 with four N application rates: 300 (N300), 225 (N225), 150 (N150), and 0 Kg ha−1 (N0) in the Northwest of China. GY was assessed by measuring the photosynthetic capacity and root system (root volume, surface area, length density and distribution). Grain yield decreased by −3%, 7.7%, and 21.9% when the N application rates decreased by 25%, 50%, and 100% from 300 Kg ha−1. We found that yield reduction driven by N reduction was primarily due to decreased radiation use efficiency (RUE) and WUE instead of intercepted photosynthetically active radiation and evapotranspiration. In the N225 treatment, GY, WUE, and RUE were not significantly reduced, or in some cases, were greater than those of the N300 treatment. This pattern was also observed with relevant photosynthetic and root attributes (i.e., high net photosynthetic rate, stomatal conductance, and root weight, as well as deep root distribution). Our results suggest that application of N at 225 Kg ha−1 can increased yield by improving the RUE, WUE, and NUE in semi-arid regions.