Despite decades of work implementing land-based best management practices (BMPs), the amount of nitrogen (N) moving down streams remains unacceptable in many watersheds. During this time, stream restoration has become increasingly popular, yet efforts to quantify the N-removal benefits of restoration are only just beginning. Because restoration practitioners cannot put projects on hold until more data are available, they are asking scientists to use the best available knowledge to provide advice that will maximize chances of reducing the downstream flux of N. We propose a framework for prioritizing sites that involves identification of N sources, mode of delivery to streams, timing of delivery, and stream size. Initial screening of sites should be based on knowledge of the magnitude of potential N loads which is largely controlled by land use. Next, the relative rates of N transmission to streams in these regions should be estimated based on knowledge of climate, hydrologic conditions, and subsurface litho-geochemistry. Thus, on a regional basis, managers could begin the prioritization process by overlaying hydrogeologic maps and physiographic data with land use data. Within a priority region, small streams (1^st to 3^rd order) that export the majority of N during low flows offer the greatest restoration opportunities and should be targeted. Restoration designs should focus on optimizing in-stream carbon availability, increasing contact between the water and benthos, and improving connections between streams and adjacent terrestrial environments. A multi-pronged approach that combines stream restoration with land-based BMPs offers the greatest opportunity to reduce N movement downstream.