A Distributed Time-Variant Gain Hydrological Model (DTVGM) based on remote sensing (RS) is proposed. The model contains several sub-models, such as snowmelt model, runoff model. It produces outputs including snow cover, evaporation, runoff, etc. All inputs for the model are derived from remote sensing data. Data from the Lhasa River basin is used in this study, including USGS-SRTM DEM, TRMM precipitation and Modis-LST. More than eight years (2001-2008) of daily hydrological data set was selected to calibrate the model. Based on the comparison of the observed and estimated runoff, the model's averaged efficiency of daily runoff simulation is over 0.6.  The error of water balance was less than 5%. Distributed modeling results are quite satisfactory. This study provided a promising approach to resolve hydrology and water resources problems in ungauged or sparsely gauged basins.