According to the temperature and precipitation data observed by 72 national meteorological stations from 1961 to 2005 and simulated by the regional climate model CCLM from 1961 to 2005 in the Yellow River Basin of Inner Mongolia, the BP neural network model is used to predict and analyze the runoff changes in the Toudaoguai hydrological station under three RCP scenarios from 2011 to 2100, so to evaluate the possible impact of future climate change on water resources in the basin. The results show that: (1) From 2011 to 2100, the temperature in the Yellow River Basin of Inner Mongolia would increase, but the precipitation would not change significantly, and the annual average runoff show a decreasing trend. Corresponding to the scenarios of RCP2.6, RCP4.5, and RCP8.5, the decrease would be by 3.6%, 2.7% and 23.4%, respectively. (2) In the future, the spring runoff would mainly increase; in summer, the trends of runoff change in different scenarios would be inconsistent; in autumn, the runoff would mainly increase before the 2050s and then decrease mainly; in winter, the runoff would mainly decrease. (3) In the future, the available water resources in the basin would decrease, and especially the contradiction between supply and demand of water resources would intensify in summer. The seasonal distribution of runoff would change, which might result in greater spring runoff.