More than one half of the total area of Mauritius Island (1844 km2) is under intensive cultivation, mostly sugarcane. Since the sugarcane industry is currently facing tremendous economic constraints, sugarcane cultivation may be diversified into other agricultural types such as vegetables, pineapple and forestry. Increasing concern about the sugarcane industry and the consequences of agricultural diversification, necessitated the application of soil loss prediction models within a GIS framework. Modelling of the potential soil loss in the Rivierre Des Anguilles catchment (RDAC) is undertaken to understand the extent to which soil erosion is affected by different land use types or agricultural systems. Although most of the RDAC is covered with sugarcane (62%), a wide range of landforms, micro-climates and soils exist, making the catchment representative of southern catchments in Mauritius. The study integrates GIS techniques with two empirical soil loss models: The Revised Universal Soil Loss Equation (RUSLE); and The Soil Loss Estimation Model of Southern Africa (SLEMSA). Both models, as well as the GIS application termed Soil Erosion Assessment using GIS (SEAGIS), are used to investigate average annual soil loss from the catchment under key management practices. Using data on soil erodibility, rain erosivity, topography and land cover, soil loss can be estimated under different management options for cropland (sugarcane, intercropped cane, vegetables, banana and tea) and natural vegetation (scrub and forest). RUSLE is additionally used to predict soil loss for the catchment under potential crop diversification scenarios including, vegetables, pineapple and forest. Using the empirical soil loss models in conjunction with a GIS, it is possible to compile soil erosion prediction maps of the RDAC under current and future conditions. Although soil loss in the catchment varies significantly, models show a similar trend in mean soil loss rates of the cropping systems. Rates are generally highest on steep slopes (>20%) with high rainfall (2400 mm) along the river valley and upper catchment area (above the 400 contour line). Predicted soil loss results, however, indicate a strong inverse relationship with vegetation cover. Very high soil loss values (more than 80 t.ha-1.yr-1) are attained under vegetables, moderate values (13 to 20 t.ha-1.yr-11) under intercropped cane, low (10 t.ha-1.yr-1) or very low (less than 2 t.ha-1.yr-1) under sugarcane, very low (4 t.ha-1.yr-1) to moderate (16 t.ha-1.yr-1) ratings under banana plantations, very low (less than 1 t. -1.yr-1) to high rates (41 t.ha-1.yr-1) under tea plantations, and low rates (less than 10 t.ha-1.yr-1) for natural vegetation. SLEMSA, however, predicts high erosion rates (27 t.ha-1.yr-1to 59 t.ha-1.yr-1) under natural vegetation, since the model is not developed for use in natural conditions. Crop diversification will have a considerable influence on soil erosion. RUSLE predicts a mean soil loss of 42 t.ha-1.yr-1, 20 t.ha-1.yr-1, and 0.2 t.ha-1.yr-1 under vegetables, pineapple, and forest, respectively. When compared to current conditions, the mean soil loss for the catchment will double under pineapple (increase by 100%), and quadruple under vegetables (increase by 300%). Results indicate that no appreciable erosion damage will occur in the RDAC if converted to forested land. Results provide considerable information regarding soil loss under potential land use The study also improves the understanding of factors governing erosion in Mauritius, which is important in the targeting of research and soil conservation efforts. Landowners and the government can use results to promote farming systems that do not degrade land resources.