Managed Aquifer Recharge (MAR) is promoted as an attractive technique to meet growing water demands. An impediment to MAR applications, where oxygenated water is recharged into anoxic aquifers, is the potential mobilization of trace metals (e.g., arsenic). While conceptual models for arsenic transport under such circumstances exist, they are generally not rigorously evaluated through numerical modeling, especially at field-scale. In this work, geochemical data from an injection experiment in The Netherlands, where the introduction of oxygenated water into an anoxic aquifer mobilized arsenic, was used to develop and evaluate conceptual and numerical models of arsenic release and attenuation under field-scale conditions. Initially, a groundwater flow and nonreactive transport model was developed. Subsequent reactive transport simulations focused on the description of the temporal and spatial evolution of the redox zonation. The calibrated model was then used to study and quantify the transport of arsenic. In the model that best reproduced field observations, the fate of arsenic was simulated by (i) release via codissolution of arsenopyrite, stoichiometrically linked to pyrite oxidation, (ii) kinetically controlled oxidation of dissolved As(III) to As(V), and (iii) As adsorption via surface complexation on neo-precipitated iron oxides.