Increases in nutrient loading are an on-going problem in many watersheds including the Chesapeake Bay watershed. Best management practices (BMPs) are used to reduce loadings to watershed tributaries. However, budget constraints often make determining the most effective nutrient load reducing BMP set a complex task. This study demonstrates a systematic approach to the BMP planning process by using deterministic optimization methods to find optimal BMP plans given a set of pre-existing loading and budget constraints. A geographic information system is used as a spatial computation and user interface tool to help perform these optimizations. The model that is developed and presented is used in a series of simple experiments to determine characteristics of optimal BMP plans. These experiments showed that optimal BMP plans can be more effectively developed by relying on numerical optimization methods than by using decisions made intuitively by experts.