Soil erosion is a major problem confronting land and water resources in many parts of the world and the spatial extent should be assessed and continually monitored. The combination of existing erosion models and remote sensing techniques within a Geographical Information System framework is commonly utilized for erosion risk assessment. In most countries, however, especially in developing countries such as South Africa, there is still an absence of standardized methodological frameworks that deliver comparable results across large areas as a baseline for regional scale monitoring. Assessment at the regional scale is often problematic due to spatial variability of the factors controlling erosion and the lack of input and validation data. Due to limitations of scale at which techniques can be applied and processes assessed, this study implemented a multi-process and multi-scale approach to support establishment of a methodological framework for South African conditions. The approach includes assessment of (i) sheet-rill erosion at a national scale based on the principles and components defined in the (Revised) Universal Soil Loss Equation, (ii) gully erosion in a large catchment located in the Eastern Cape Province by integrating eleven important factors into a GIS, and (iii) sediment migration for a research catchment near Wartburg in KwaZulu-Natal by means of the Soil and Water Assessment Tool. Case Study i illustrates that 20% (26 million ha) of South African land is classified as having a moderate to severe actual erosion risk (emphasizing sheet-rill erosion) and describes the challenges to be overcome in assessment at this scale. Case Study ii identifies severe gully erosion affecting an area of approximately 5 273 ha in the large catchment (Tsitsa valley) of the Eastern Cape Province and highlights gully factors likely to emerge as dominant between continuous gullies and discontinuous gullies. Case Study iii illustrates that a cabbage plot in the upper reaches of a research catchment near Wartburg is a significant sediment source, but is counterbalanced by sinks (river channel and farm dams) downstream. Model assumptions affecting outputs in the context of connectivity between sources and sinks are described. The factor-based nature of this multi-process and -scale approach allowed scrutiny of the role of the main factors in contributing to erosion risk. A combination of poor vegetation cover and susceptible parent material-soil associations are confirmed as the overriding factors in South Africa, and not topography and rainfall as frequently determined in the USA and Europe. A methodological framework with three hierarchical levels is then presented for South Africa. The framework illustrates the most feasible erosion assessment techniques and input datasets for which sufficient spatial information exists, and emphasizes simplicity required for application at a regional scale with proper incorporation of the most important factors. The framework is not interpreted as a single assessment technique but rather as an approach that guides the selection of appropriate techniques and datasets according to the complexity of the erosion processes and scale dependency. It is useful in determining the relative impact of different land use and management scenarios, as well as for comparative purposes under possible future climate change scenarios.