Concern for nonpoint source pollution from rangelands has increased the need to monitor and predict amounts of soil erosion that may enter streams from adjacent rangelands. This study was undertaken to evaluate a refined surface cover subfactor (RSC) developed for the revised universal soil loss equation (RUSLE) to simulate soil loss from rangelands as affected by various kinds and amounts of surface cover. In addition, sensitivity analysis indicated which variables most influenced erosion from a sagebrushgrass rangeland (Johnson and Gordon 1988). Evaluation of the RSC was done by comparing original RUSLE estimates of soil loss with those of RUSLE where the surface cover subfactor (SC) had been replaced with RSC. Estimated soil loss from both simulations were compared with field measurements of soil loss using a simple regression technique. Refined RUSLE predictions of soil loss (r2 = 0.81 and 0.50 for dry and moist conditions) were considerably better than those obtained with the original RUSLE model (r2 = 0.67 and 0.14 for dry and moist conditions). The refined RUSLE was better at describing Johnson and Gordon (1988) erosion data than was the original RUSLE model. Use of RSC in the RUSLE model may increase its accuracy, but the model still underpredicts the actual amount of soil loss.