Soil erosion is a serious concern around the world because of its detrimental effect on agricultural productivity and reservoir storage capacity, which is mostly caused by improper land management. Effective land management entails expensive soil conservation measures which are not affordable in developing countries, especially in large watersheds. Thus, identifying the most vulnerable areas for sediment yield and executing remedial measures for those areas is critical to lower costs and maintain effectiveness. The Soil and Water Assessment Tool (SWAT) model is one of the commonly used models to identify weak zones for erosion from smaller units (hydrologic response units). In this study, the watershed of Tekeze was divided into 34 subbasins and 158 hydrologic response units, and the areas most vulnerable for sediment production were identified using the SWAT model. The model was evaluated using SWAT-CUP SUFI 2 algorithms by comparing predicted streamflow and sediment yield in the Tekeze basin with observed values. For streamflow, R2 = 0.76, 0.85, NS = 0.76, 0.69, PBIAS=-8.1, 13.4, and for sediment, R2 = 0.8, 0.66, NS =0.8, 0.62, PBIAS=-4.1, 4.1 for calibration and validation, respectively. Results showed a large variation in sediment yield from one subbasin to the other. Out of 34 subbasins, four had a very high sediment yield (21 to 39 t/ha/yr), two had high (16 to 20 t/ha/yr), one had medium (11 to 15 t/ha/yr), four had low (6 to 10 t/ha/yr) and the rest had a very low sediment yield (0 to 5 t/ha/yr). One subbasin had quite a high SDR(> 0.45), and the other three had a high SDR (0.45-0.32). The overall result sediment delivery ratio indicated that subbasins 1, 3, 5, 14, 17, and 28 required great attention.
