This paper examines the utilization of surface temperature as a variable which can be assimilated in off-line land surface hydrological models. The connection between the surface temperature and evapotranspiration is utilized in making adjustments to the model-computed surface soil moisture. This adjustment is a function of the difference between the model-computed and the observed surface temperature. Comparisons between the model-computed and satellite-observed surface temperatures have been carried out. The assimilation of surface temperature is carried out twice a day (corresponding to the A.M. and P.M. overpass of the NOAA 10) over the Red-Arkansas basin in the southwestern United States (31 ¿50'N--36 ¿N, 94 ¿30'W--104 ¿30'W) for a period of 1 year (August 1987 to July 1988). The soil moisture estimates resulting from the assimilation of surface temperature have a closer agreement with the values derived from the special sensor microwave imager than those from simulations without surface temperature assimilation. Assimilation reduces the effect of errors in precipitation and/or shortwave radiation on simulated soil moistures. ¿ 2000 American Geophysical Union