As one of the main environmental problems of global change, the increase of atmospheric nitrogen deposition has aroused widespread concern. Continuous nitrogen deposition has an important impact on the composition, structure and function of grassland ecosystems. In order to deeply understand the response of community biomass and root-shoot ratio to nitrogen deposition in the process of grassland restoration, this study took 3 natural grasslands with different recovery years (early 12 years, middle 28 years, and late 37 years) on the Loess Plateau as the research object. Six nitrogen addition levels, CK (0), N1(2.34 g m^-2 a^-1),N2(4.67g m^-2a^-1),N3(9.34g m^-2a^-1),N4(18.68g m^-2a^-1),N5(37.35g m^-2a^-1)were used to measure the aboveground biomass, underground biomass, and total biomass of grassland communities, and the root-shoot ratio and nitrogen response efficiency (NRE). The results showed that:(1) The aboveground biomass was the largest in the middle stage of restoration. With the increase of nitrogen addition gradient, the aboveground biomass showed an insignificant upward trend in the early and late stages of restoration, a non-linear positive response to nitrogen addition (ΔNRE>0), an insignificant downward trend in the middle stages of restoration, and a non-linear negative response to nitrogen addition (ΔNRE<0). (2) The underground biomass of the community had no significant response to nitrogen addition, and the total biomass was significantly different from that of the control only at the N4 addition level in the later stage of restoration. (3) At the early stage of restoration, the root-shoot ratio at the N3 addition level was significantly higher than that of the control and other nitrogen addition levels, while the other restoration years had no significant response to nitrogen addition. In conclusion, the responses of aboveground, underground, total biomass and root-shoot ratio of grassland communities with different restoration years on the Loess Plateau to nitrogen addition were analyzed and compared. We suggest that pilot experiments should be carried out in this region to implement adaptive grassland management, such as biennial cutting or light grazing (2 sheep/hm²), to explore more scientific and effective management measures to achieve systematic restoration of grassland, so as to meet the changes of ecosystem capacity and social needs.