Elevated levels of selenium (Se) in aqueous environments can harm aquatic life and endanger livestock and human health. Although Se occurs naturally in the rocks and soils of many alluvial aquifers, mining and agricultural activities can increase its rate of mobilization and transport to surface waters. Attention is given here to regions where nonpoint source return flows from irrigated lands carry pollutant loads to aquifers and streams, contributing to concentrations that violate regulatory and performance standards. Of particular concern is the heightened level and mobilization of Se influenced by nitrate (NO₃), a harmful pollutant in its own right. We present a numerical model that simulates the reactive transport of Se and nitrogen (N) species in a coupled groundwater-surface water system. Building upon a conceptual model that incorporates the major processes affecting Se and NO₃ transport in an irrigated watershed, the model links the finite-difference models MODFLOW, UZF-RT3D, and OTIS, to simulate flow and reactive transport of multiple chemical species in both the aquifer and a stream network, with mass exchange between the two. The capability of the new model is showcased by calibration, testing, and application to a 500 km² region in Colorado’s Lower Arkansas River Valley using a rich data set gathered over a 10-yr period. Simulation of spatial and temporal distributions of Se concentration reveals conditions that exceed standards in groundwater for approximately 20% of the area. For the Arkansas River, standards are exceeded by 290%–450%. Simulation indicates that river concentrations of NO₃ alone are near the current interim standard for the total of all dissolved N species. These results indicate the need for future use of the developed model to investigate the prospects for land and water best management practices to decrease pollutant levels.