Agricultural non-point source phosphorus (P) loading to tributaries is a significant mechanism for the eutrophication of Lake Simcoe. This thesis is an investigation of P chemistry in agricultural soils of the Maskinonge river watershed to identify critical source areas where remedial efforts can be targeted. Soil assemblage models were developed to predict P adsorption to charged soil minerals; proton- and P-binding constants of four clay samples were determined by potentiometric titration and batch adsorption experiments respectively. Binding constants of goethite were obtained from the literature. The models were translated to the landscape by means of the digitized legacy soil map of York County. The model-predicted P adsorption capacities of the mapped soils were categorized into a five-level P adsorption/risk index. The index ratings were applied to the soil map, creating a soil P sensitivity map for the Maskinonge watershed.