The source area of the Yellow River is an important part of the Yellow River Basin, and its runoff change affects the water resources and ecosystem security of the whole basin. Using the runoff data from 1976 to 2014, meteorological data, digital elevation model (DEM), land use, soil and meteorological data of three future scenarios in eight models of the 6th coupled model inter-comparison project (CMIP6) from 2021 to 2100, and based on the soil and water assessment tool (SWAT) hydrological model, this study simulated, projected, and analyzed the future runoff and variation of main hydrological stations in the source area of the Yellow River. The results show that: (1) The SWAT model has good applicability in historical runoff simulation in the source area of the Yellow River. The uncertainty of runoff simulation is small, and the simulated values are close to the measured values. (2) Parameter sensitivity analysis showed that 27 hydrological parameters have a certain impact on runoff simulation. Among them, soil evaporation compensation factor, the number of SCS (Soil Conservation Service) runoff curves under humid condition II, and shallow groundwater runoff coefficient are highly sensitive, and runoff is greatly affected by land surface evapotranspiration, underlying surface, and precipitation. (3) Precipitation is the main factor affecting future runoff. Under the two future scenarios of SSP126 and SSP245, the annual average discharge of Jimai, Maqu, and Tangnaihai hydrological stations shows an increasing trend in the two periods from 2021 to 2100, while under the SSP585 scenario, it shows an increasing trend from 2021 to 2060 and a decreasing trend from 2061 to 2100. Relative to 1976-2014, the annual average discharge at Tangnaohai and Maqu stations in the near future (2021-2060) increases the least under the SSP585 scenario and the most under the SSP126 scenario, and at Jimai station it increases the most under the SSP245 scenario and the least under the SSP126 scenario. Annual average discharge at three hydrological stations in the far future (2061-2100) increases the least under the SSP585 scenario and the most under the SSP245 scenario, except for Jimai station, which has the lowest increase in the SSP126 scenario. The research results can provide important scientific basis and theoretical support for water resources management, flood control and water storage, as well as ecological environment protection in the Yellow River Basin.