This study investigated the effectiveness of vegetative filter strips (VFS) in removing phosphorus from surface runoff. Dissolved and particulate nutrients were treated separately due to differing transport and removal mechanisms. Nutrient transport in VFS appears to be a function of runoff rate, concentration and size distribution of suspended solids, and biological factors that influence hydrologic and chemical processes in filter strips. Three sets of experimental field plots were constructed to simulate VFS. Each set consisted of three plots containing sediment and nutrient source areas and 0.0, 4.6, or 9.1 m grass filter strips. Artificial rainfall was applied to the plots, and surface runoff, soil, and plant material samples were collected and physically and chemically analyzed. The VFS reduced surface runoff, suspended solids, and phosphorus losses. Most removal of sediment and phosphorus was accomplished in the first few meters of the VFS. The filter strips did not remove phosphorus as effectively as sediment, due to their ineffectiveness for filtering dissolved phosphorus and sediment-bound phosphorus associated with fine particles. The VFS often increased orthophosphorus losses in surface runoff. Laboratory batch experiments of phosph~rus desorption reaction suggested that plant residues, living plant canopy, and soil components of the strips could release dissolved phosphorus to surface runoff. A modified Elovich equation and a diffusion-control model were used to describe the phosphorus release from the plant and soil materials. A computer model, GRAPH, was developed to simulate phosphorus transport in VFS by incorporating phosphorus transport submodels into the VFS model in SEDIMOT II, a stormwater and sediment transport model. The model considers the effects of advection processes, infiltration, biological uptake, phosphorus desorption from the soil surface to runoff, the adsorption of dissolved phosphorus to suspended solids in runoff, and the effects of dynamic changes in the sediment size distribution on chemical transport. GRAPH was verified using the results of the physical plot simulations. The model's predictions and observed phosphorus transport compared favorably. Sensitivity analysis suggested that sediment and phosphorus removal was sensitive to the input parameters in the order: filter length and width, grass spacing, and filter slope and surface roughness. Increased filter width and length and aboveground biomass increased orthophosphorus loss from VFS. / Ph. D.