As nutrients and sediment in agricultural watersheds continue to degrade water quality, attention is increasingly given to reverse auctions to cost-effectively address these pollutants. Typically, reverse auctions include a selection process which depends on both the monetary bid and a ranking of the environmental benefit, where the latter is often approximated using simple models, such as the Universal Soil Loss Equation (USLE). When the environmental objective is to improve water quality, the cost-effectiveness of such ranking methods cannot always be assured since simple models may poorly approximate the effects on downstream water quality. In this paper, we introduce an alternative reverse auction approach that takes advantage of richer watershed process models and optimization tools that are now much more commonly available. This “improved” reverse auction allows decision-makers to better consider the cost-effective assignment of conservation practices and to address water quality or other environmental objectives. In a spatially detailed simulation, we demonstrate how this approach can improve the design of a reverse auction for the Raccoon River Watershed in Iowa, and estimate the potential gains from using the simulation-optimization approach relative to simpler ranking methods for selecting bids. We also point out that simple bid ranking schemes may not yield sufficient nutrient reductions to achieve water quality goals but bids are easily selected to achieve any feasible water quality improvement in the “improved” auction process.