Considering the hydrological calculation and prediction of ungauged basins, a distributed hydrological model, the THREW model, was adopted to study the upper basin of the Brahmaputra River from the Nuxia Hydrological Station. Surface meteorological observation, remote sensing of vegetation cover, snow cover area, and section hydrological monitoring data were used to calibrate the model, and the latest CMIP5 data were used to estimate the runoff evolution. The results showed that, in the upper basin of the Brahmaputra River from the Nuxia Hydrological Station, the Nash-Sutcliffe efficiency coefficient of monthly runoff simulation of the calibration period from 1991 to 1995 was 0.75, the coefficient of monthly runoff simulation of the validation period from 1996 to 2000 was 0.76, and the runoff markedly increased when the CO_2 discharge in the AR5 of IPCC was maximal.