Runoff from animal feeding operations is a major source of water pollution. Vegetative filter strips (VFS) are effective ways to reduce nonpoint source pollution. In this study, vegetative filter strips with different designs and in climatic and management conditions of North Dakota were evaluated. Runoff samples were collected from inflow (before entering VFS) and outflow (after exiting the VFS) locations using automatic samplers. Collected samples were analyzed for solids and nutrients. It was observed that the transport reductions by VFS were ranged from very low to up to 100%. However, soluble nutrients were not as effectively removed as sediment and sediment bound nutrients. Filter with longer length was more effective in reducing transport of sediments and nutrients. Antecedent soil moisture condition had an important effect on VFS performance. An attempt was made by varying the VFS soil pH in a broader range to investigate effect of pH on reducing transport of soluble nutrients from manure borne runoff. Soil was treated with calcium carbonate to adjust pH at different levels. Treated soil was packed into galvanized iron boxes and seeded with grasses to simulate vegetative filter strips. Runoff experiments were conducted with manure solution and inflow, outflow, and leachate samples were collected. Samples were analyzed for sediment and nutrients. It was observed that the soluble nutrients transport was influenced by the pH, and higher ortho-P transport reduction was observed in higher pH. Leaching of NO3-N at higher pH was observed, indicating potential of groundwater pollution from the soil with higher pH. Using calcium carbonate to increase soil pH and thereby reducing transport of soluble nutrient could increase VFS performance. To aid VFS design and evaluation, a model was developed to predict trapping efficiency of sediment, sediment bound P, and dissolved P from VFS. Two procedures were coded into FORTRAN and added into existing VFSMOD model. The model was calibrated and validated using field data. Due to limited data points and difficulties in measuring runoff volume, the model appeared to be under or over predicting. In future, model predictability can be improved by accurately measuring runoff volume and carefully selecting input parameters.