This PhD study assessed the impacts of land use/cover changes over the past few decades on runoff and soil loss in a watershed in Northwestern Ethiopia. In chapter 2, land use/cover changes for the past 5 decades were generated from remotely sensed data after the input data were reconciled for the differences in spatial scales. The remotely sensed data analysis was complimented by ground truth data through field survey and interviews with elders that have observed the land use/cover changes in the study area for the considered time period. Results indicate that woody vegetation cover in the study area has tremendously decreased from the 1950s to 2010. Most of the changes took place between the 1970s and 1980s. The area used for settlement significantly increased from the 1950s to 2010. Population pressure and land use policies accounted for the changes in land use/cover. In Chapter 3, high temporal resolution measurement of turbidity and discharge data, accompanied by sediment-discharge hysteresis analysis enabled us to estimate the seasonal variation and sources of sediment in the study area. While agricultural areas and hill slopes contribute most of the sediment yield in the study area, most of the sediment loss occurs in the high rainfall months between June and mid-August when the agricultural areas have very little vegetation cover. The following high rainfall months of the year between mid-August and the end of September contributed only a small portion of the total yearly sediment yield as the agricultural areas are completely covered with vegetation within this time period. 25 t yr-1ha-1 rate of suspended sediment transport was estimated in the study area. In chapter 4, the decadal trends in runoff and soil loss for changes in land use/cover were simulated using the Annual Agricultural non-point source (AnnAGNPS) model. The model was calibrated with measured yearly runoff and soil loss data and then with daily data for one year followed by validation on the following year. The model predicted runoff and soil loss well with good correlation coefficients between measured and modeled values and satisfactory Nash-Sutcliffe coefficients. Model simulations for changes in land use/cover over the past 5 decades produced runoff and soil loss maps for that time period. Simulation results indicated a tremendous increase in soil loss and runoff from the 1950s to 2010. The modeling output enabled delineation of the areas that were at high erosion risk throughout the entire past 5 decades, as well as the current high risk erosion areas, for planning appropriate soil and water conservation measures that may reduce soil loss and runoff. In chapter 5, selected soil and water conservation measures were simulated with the calibrated and validated AnnAGNPS model in chapter 4. Simulation results indicate that soil loss in the study area can be reduced by up to 88% and runoff by up to 22% when combinations of reforestation, contour farming and terracing were applied, each of them on areas with different levels of erosion risk. Chapter 6 summarizes the findings and conclusions of this study.