The potential utilization of switchgrass (Panicum virgatum L.) as a cellulosic energy crop was evaluated as a component of a projected future national network of biorefineries designed to increase national reliance on renewable energy from American farms. Empirical data on yields of switchgrass from a network of experimental plots were coupled with data on switchgrass physiology and switchgrass breeding progress to provide reasonable expectations for rates of improvement over current yields. Historical breeding success with maize (Zea mays L.) was found to provide a reasonable model for projected linear rates of yield improvement of switchgrass based on documented progress to date. A physiologically based crop production model, ALMANAC, and an econometric model, POLYSYS, were utilized to estimate variability in switchgrass yield and resource utilization across the eastern two-thirds of the United States. ALMANAC provided yield estimates across 27 regional soil types and 13 years of weather data to estimate variability in relative rates of production and water use between switchgrass and maize. Current and future yield projections were used with POLYSYS to forecast rates of adaptation and economic impacts on regional agricultural markets. Significant positive impacts on US markets, including significant increases in farm income and significant reduction in the need for government subsidies, were projected. This was based on expected technological progress in developing biorefineries that will significantly increase national energy self-sufficiency by producing feed protein, transportation fuel, and electrical power from cellulosic feedstocks. © 2006 Elsevier Inc. All rights reserved.