Long-term information on switchgrass (Panicum virgatum L.) as a biomass energy crop grown on marginally saline soil and the associated impacts on soil carbon (C) and nitrogen (N) dynamics, greenhouse gas (GHG) emissions, and best management practices (BMPs) are limited. In this study, we employed the DAYCENT model, based on a 4-year switchgrass field experiment, to evaluate the long-term biomass yield potential and environmental impacts, and further to develop BMPs for switchgrass in a semi-arid region. The model showed that long-term (14-year) annual mean biomass yields were 9.6 and 5.2 Mg ha(-1) for irrigated and rainfed switchgrass systems, respectively. The simulated biomass yields correlated well with field-measured biomass with r(2) values of 0.99 and 0.89 for irrigated and rainfed systems, respectively. Soil organic carbon (SOC) and soil total nitrogen (STN) accumulated rapidly after switchgrass establishment, with mean accrual rates of 0.99-1.13 Mg C ha(-1) yr(-1) and 0.04-0.08 Mg N ha(-1) yr(-1), respectively. Based on the outputs of numerous long-term model simulations with variable irrigation water supplies and N rates, the irrigation regime and N rate with the highest yield to input ratio were chosen as BMPs. The DAYCENT model predicted-BMP was irrigating every 14 days at 70% potential evapotranspiration combined with an N rate of 67 kg ha(-1) yr(-1). Switchgrass established and produced biomass reasonably well in this semi-arid region; however, appropriate irrigation and N fertilization were needed for optimal biomass yield. Switchgrass had a great potential to sequester C into soils with low N2O emissions while supplying significant quantities of biomass for biofuel synthesis. (C) 2015 Elsevier Ltd. All rights reserved.