Release and spreading of permanganate (MnO4-) in the well-based controlled-release potassium permanganate (KMnO 4 ) barrier system (CRP system) was investigated by conducting column release tests, model simulations, soil oxidant demand (SOD) analyses, and pilot-scale flow-tank experiments. A large flow tank (L×W×D=8m×4m×3m) was constructed. Pilot-scale CRP pellets (OD×L=0.05m×1.5m; n=110) were manufactured by mixing ∼198kg of KMnO 4 powders with paraffin wax and silica sands in cylindrical moulds. The CRP system (L×W×D=3m×4m×1.5m) comprising 110 delivery wells in three discrete barriers was constructed in the flow tank. Natural sands (organic carbon content=0.18%; SOD=3.7–11gMnO4-kg −1 ) were used as porous media. Column release tests and model simulations indicated that the CRP system could continuously release MnO4- over several years, with slowly decreasing release rates of 2.5kgd −1 (day one), 109gd −1 (day 100), 58gd −1 (year one), 22gd −1 (year five), and 12gd −1 (year 10). Mean MnO4- concentrations within the CRP system ranged from 0.5 to 6mgl −1 during the 42 days of testing period. The continuously releasing MnO4- was gradually removed by SOD limiting the length of MnO4- zone in the porous media. These data suggested that the CRP system could create persistent and confined oxidation zone in the subsurface. Through development of advanced tools for describing agent transport and facilitating lateral agent spreading, the CRP system could provide new approach for long-term in situ treatment of contaminant plumes in groundwater.