Estimates of soil loss are important to issues of land and water management. Much of the soil loss information in erosion control is based on the use of the Revised Universal Soil Loss Equation (RUSLE). This study assessed use of RUSLE as a soil loss predictor on small drainage areas using sediment data. Toward this, twenty drainage areas covering the first order waterways were selected in the semi-arid Tahamchai Watershed, NW Zanjan, Iran. Sediment yield was measured using sediment volume deposited behind the check dam at the outlet of each drainage area for a 16-year study period (1995-2011). Soil erosion factors including rainfall erosivity, soil erodibility, slope, vegetation cover and conservation practices were determined for each drainage area. Soil loss was estimated using the RUSLE model integrated with RS and GIS techniques based on the maps of rainfall erosivity, soil erodibility, slope, cover crop, and conservation practices factors. Based on the results, mean annual estimated soil loss was 43.68 Mg ha^-1 year^-1 which was 6.58 times bigger than the mean annual sediment yield (8.67 Mg.ha^-1.year^-1) in the area. The model efficiency was -6.83, indicating the inability of the model to correctly predict net soil loss in the small drainage areas. The estimation error was directly related to overestimation of soil erosion factors in the area. It seems soil erodibility, slope and conservation factors to have the exaggerated estimates in the area on one hand, and some other variables such as slope shape and waterway characteristics to be also effective in controlling the soil loss in the area on the other hand. This study is an important step forward in conducting more accurate erosion evaluation of the drainage areas using the RUSLE.