Measuring soil moisture turns out to be crucial in watershed management. This study presents soil moisture prediction using RADRASAT-1 Synthetic Aperture Radar (SAR) satellite imagery collected in the Choke Canyon Reservoir Watershed (CCRW) in April 2004. Essential radiometric and geometric calibrations to correct the SAR imagery were performed with the aid of Corner Reflectors (CRs). The sensor data obtained after the installation of the corner reflectors in April 2004 showed better spatial accuracy, and consequently improves the correlation between the radar backscatter signals, σ 0 , and the Volumetric Moisture Content (VMC) of the soil in CCRW. Three prediction models were developed for soil moisture projection, which include simple linear regression, multiple linear regression, and genetic programming models. It found that the genetic programming model exhibits overall advantage of soil moisture estimation.