Non-point source sediment and nutrient runoff from upstream agricultural production is known to impair downstream ecosystem functions and services. Despite adoption of agricultural best management practices (BMPs) by some farmers, there are still many cropland parcels with a high need for additional conservation treatment to reduce soil erosions and runoffs. Despite the well documented environmental benefits of BMPs, significant uncertainty still remains regarding the effectiveness of policies that promote these practices. The realization of environmental benefits through improved BMPs is most constrained by our limited understanding of how farmers respond to policies and the differences in responses across different types of farmers. Previous studies of agricultural management decisions are either not spatial or omiting farmer characteristics. We will improve on these approaches by combining a model of farmer behavior with a spatial model of land management across all parcels in the watershed. In this research we develop a spatially explicit behavioral model of farmers’ BMP choices that accounts for both farmers’ socioeconomic characteristics and spatial variations of land parcels. We apply this model to the Maumee River Watershed and to three BMPs – crop rotations, conservation tillage and Conservation Reserve Program enrollment. With this model, we can simulate BMP outcomes for each parcel across the entire watershed under baseline and alternative policy scenarios. The spatial behavioral model is expected to outperform a historical trend model and allows us to evaluate a broader range of potential policies in terms of their simulated impacts on farmer behavior, BMP decisions, and downstream ecosystem conditions.