Identifying areas of the hillslope that are most sensitive to soil erosion and contribute significantly to sediment yield is a primary concern in environmental protection and conservation. Therefore the ability to predict the magnitude and variability of soil erosion and sediment yield is important to catchment managers in order to select the appropriate conservation practices that keep soil erosion and sediment yield within the tolerable limits. A number of models have been developed for simulating soil erosion and sediment yield from a catchment. However, none of them are universally applicable and most of them require extensive data which are extremely costly, time consuming and sometimes not available except in research catchments. Hence it was concluded that the combined use of an empirically based soil loss model, RUSLE, Geographic Information Systems (GIS) techniques, and a Sediment Delivery Ratio (SDR) concept would be a candidate modelling tool, which would be a compromise between the advantages of simplicity, data availability, the complex spatial variability of hydrological and geomorphological characteristics of a catchment and the economic limitation of field data measurements in sediment yield studies. Such a modelling tool was developed in this research and was able to identify sediment source areas and predict annual sediment yield from catchments. Data from the Henley catchment, South Africa have been used for demonstrating the potential use of the model in soil erosion and sediment yield studies. Arcview GIS grid functions were used to define the flow direction, accumulation, pathways, and velocity in a catchment as a function of topography and land use and to describe spatially variable input and output information. In addition the Arcview GIS grid function was used to discretise the catchment into hydrologically homogeneous grid cells to capture the catchment heterogeneity. The gross soil erosion in each cell was calculated using the soil loss model RUSLE while a distributed topography based SDR parameter was used to determine the mass of eroded sediment that would be transported to the nearest stream and ultimately to the catchment outlet. The average annual soil loss and sediment yield values were 26 t. ha-1.yr -1 and 1.6 t. ha-1.yr -1 respectively. High soil erosion and sediment yield rates are evident in the residential and agricultural areas, which are characterised by degradation due to overgrazing and traditional and peri-urban settlements with mixed crops. The average annual SDR value was 0.19 for the Henley catchment and large SDR values are associated with areas adjacent to the channel system. This can be explained by recognizing that the SDR is significantly influenced by characteristics of the drainage system. Comparison of event based simulations of sediment yields to those estimated from measurements demonstrated that the proposed model predictions ranged between 13 % and 60 % of the measured estimates, consistently over predicting. This is because the SDR component of the model is developed as a mean annual parameter, assuming that over a long period a stream system must intimately transport all the sediments delivered to it. Hence the channel network sediment delivery parameters would have to be considered at short temporal scales. Comparing the results of the model prediction against other sediment modelling techniques in South Africa demonstrated the usefulness of the model as an effective catchment management tool. The model has advantages over these other techniques since it includes a distributed grid based component, which enables the identification of sediment source areas in the catchment. The sensitivity analysis shows that the model was highly sensitive to parameters derived from topography and land use of the catchment. Future research with the model should include further testing and analysis of its components on different catchments. The topography based SDR concept which is a key component in sediment routing for prediction of either long term average sediment yield or isolated storm event simulation from a catchment warrants specific attention. Effort in future should focus on identifying parameters which affect the sediment delivery within a catchment. This may be achieved by incorporating processes describing the movement of sediments in the channel network of the catchment. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2004.