Thesis (MSc. Agriculture (Soil Science)) -- University of Limpopo, 2014 / Phosphorus (P) runoff from agricultural sites and the subsequent loading into surface water bodies contribute to eutrophication. Environmental concerns associated with P loading in soil have motivated the need for the development of a proper tool that will allow farmers to identify agricultural areas or management practices that have the greatest potential to accelerate eutrophication. The objective of the study was to determine the spatial variability of soil test P, soil loss potential of the farm, P application rate and methods, and map P runoff risk across the field. This study was conducted in Vierfontein Boerdery in Kriel, Mpumalanga province, South Africa (longitude 29.11258833 and latitude -26.27104340). The field was under dryland cultivation and planted to yellow maize that was rotated with soybeans. Soil samples were taken at georeferenced locations in a 100 x 100 m grid for soil analysis. Spatial layers of soil P distribution, soil loss potential as well as application rate and method were created in ArcGIS software. These layers were used as input factors in a P index model to identify areas in the farm that are vulnerable to P runoff. Results indicated a variation in soil test P. Although soil test P variation was not statistically different at P≤0.05, variation had both agronomic and environmental implications. This variation could be attributed to differences in site-specific conditions and management practices. Furthermore, soil loss potential across the study site predicted by the Revised Universal Soil Loss Equation (RUSLE) showed variation with a range of 3-15 tons/ha/yr. This variation was attributed to differences in topographic variations in the study site. There is a need for best management practices that control soil erosion to minimize P runoff into water bodies. KEYWORDS: Eutrophication, Geographic Information System, Phosphorus best management practises, Phosphorus runoff index, Soil erosion, Site-specific management.