Controlled-release, semi-passive reactive barrier systems have been recently developed as a long-term treatment option for controlling the spread of contaminant plumes in groundwater. This paper describes a new computer code, and applies it to study coupled processes of solute release, reaction, and mass transport in an in situ remediation scheme using the controlled release of potassium permanganate. Confidence with the modeling approach was developed by model verifications and simulating results of a pilot-scale test-cell experiment. Sensitivity analyses indicated the possibilities of treatment inefficiencies due to inability of transverse dispersion to mix the permanganate (MnO(4)(-)) within the zone of reaction, fluctuations in source strength due to variations in flow velocity, and the small length of treatment zone due to strong soil utilization of MnO(4)(-). Although problems associated with the fluctuating source strength and strong soil utilization can be addressed by optimizing the release rate, the inefficiency of transverse dispersion to create mixing could pose a serious limitation. Through a series of model simulations, a system of injection/withdrawal wells in a doublet arrangement was developed to facilitate lateral spreading and mixing of MnO(4)(-). A well-mixed, stable MnO(4)(-) zone with predetermined size (DxL=8m x 2m) and concentration ranges (1.5-20 mg l(-1)) was created by four 1-day injection/withdrawal pumping periods over 24 d. This type of mixing zone may persist for many years with periodic well mixing and replacements of exhausted controlled-release forms. Coupled use of the generalized code with field hydrologic data will help to optimize the design and operation of controlled-release systems in practice.