Agricultural nitrogen losses are the major contributor to nitrogen loads in the Mississippi River, and consequently, to the existence of a hypoxic, or "dead", zone in the Gulf of Mexico. Focusing on two small agricultural watersheds in southwestern Minnesota, simulation results from the Agricultural Drainage And Pesticide Management (ADAPT) model were combined with a linear-optimization model to evaluate the environmental and economic impact of alternative land-use policies for reducing nitrogen losses. Of particular importance was the study's explicit focus on agricultural subsurface (tile) drainage, which has been identified as the major pathway for agricultural nitrogen losses in the upper Midwest, and the use of drainage-focused abatement policies. Results indicate that tile-drained land plays a key role in nitrogen abatement, and that a combined policy of nutrient management on tile-drained land and retirement of non-drained land is a cost-effective means of achieving a 20- or 30-percent nitrogen-abatement goal. Results also indicate that although it is cost-effective to abate on tile-drained land, it is not cost-effective to undertake policies that plug or remove tile drains from the landscape, regardless of whether the land would be retired or kept in production. Therefore, results imply that although tile-drained land is a major source of nitrogen lost to waterways, it is not cost-effective to remove the land from production or to remove the drainage from the land. Because of its value to agricultural production, it is better to keep tile-drained land in production under nutrient management and focus retirement policies on relatively less-productive, non-drained acres.