The Tibetan Plateau plays an important role in global atmospheric circulation with its complex terrain. The downward surface short-wave radiation (DSSR) can be obtained from remote sensing data because of its sparse observations and rugged surface. In this paper, a satellite-based scheme is presented to retrieve all-sky downward surface shortwave radiation, which links a look-up table algorithm and satellite images. The look-up table for clear sky and cloudy sky was created separately using a comprehensive 1D physically based radiative transfer model SBDART to achieve a higher computational accuracy and efficiency compared to the comprehensive radiative transfer model. The estimated DSSR was validated using one year pyranometer measurements from 8 stations in the Tibetan Plateau and compared with GEWEX-SRB data with 10 spatial resolution. The result shows that the largest root mean square error was 60 W/m(2) (32%) at Guoluo station, and the least root mean square error was 13 W/m(2) (13%) at Golmud station. The bias was larger in summer and smaller in winter, which may be caused by uncertainties in the assumption from the ID radiative transfer model and MODIS cloud properties product for broken and inhomogeneous clouds. The algorithm we proposed can be applied globally without local calibration because it is independent of climate and the surface elevation. (C) 2014 Elsevier Ltd. All rights reserved.