Air sparging is a remediation technology for removing volatile organic contaminants (VOCs) from below the water table using air injection. Contaminant mass removal can be controlled by the mass transfer between liquid phases (aqueous and nonaqueous phase liquid [NAPL]) and the injected gas phase. Traditional multiphase flow numerical models ignore subgridblock-scale effects of gas channel flow on local mass transfer and assume local equilibrium between liquid and gas phases. This traditional, single-domain modeling approach tends to overpredict contaminant mass removal during sparging. A dual-domain multiphase flow approach for local interphase mass transfer may more accurately simulate contaminant removal, while still accounting for injected gas plume behavior. Mass transfer limitations are introduced through a numerical grid modification where each traditional single-domain gridblock is split into two domains: one with very strong capillary pressures that remains mostly water saturated and one with weak capillary pressures that becomes mostly gas saturated during sparging. The two domains are coupled through a first-order mass transfer expression. This simple model can then approximate the localized (subgridblock-scale) gas channel effects on mass transfer that are limited by liquid phase diffusion. In this study, data from a two-dimensional laboratory-scale experiment and a field-scale air sparging demonstration were used to test the capabilities of the dual- and single-domain approaches. It is the first time mass transfer has been simulated for air sparging at either of these scales using a multiphase flow kinetic interphase mass transfer model. Both experiments involved air sparging remediation of tetrachloroethylene, a NAPL. The single-domain local-equilibrium model resulted in an overprediction of mass removal rates in both cases. The dual-domain mass transfer approach provided a much better fit of the experimental data, and it is shown that the apparent interphase mass transfer coefficient becomes smaller with increasing scale.