Abstract: To improve the accuracy of land surface fluxes estimated by the Surface Energy Balance System (SEBS) under drought water stress conditions, the normalized difference water index (NDWI) as drought water stress information is integrated into SEBS to modify the value of parameter (kB^-1) based on linear, exponential, and S-curve equations, which decreases with the increase in water stress. The proposed scheme is calibrated and validated by meteorological and fluxes observations in 2008—2009 from the Yingke oasis station of Heihe river basin, China. Results show that the revised SEBS model can significantly improve the land surface flux estimation in arid and semi-arid regions under drought water stress conditions. The issues of underestimation of sensible heat fluxes and overestimation of latent heat fluxes with the standard SEBS can be significantly improved by the revised SEBS. Biases in estimated sensible and latent heat fluxes are reduced by 35 W/m² and 25 W/m², respectively.