Constructed wetlands are a promising technology to protect river flood plains against the impact of contaminated groundwater. They are suitable for the treatment of waters contaminated with monochlorobenzene and perchloroethene. However, the removal performance differs with the operation conditions, and generally, transferable performance data are not yet available. In this study, removal efficiencies were determined and the dominant removal processes for monochlorobenzene and perchloroethene were evaluated under various operation conditions in helophyte rhizosphere reactors. Monochlorobenzene removal was very efficient (>99%) under low carbon load (overall oxic) and moderate carbon load (overall reduced) conditions. Higher loads of easily degradable carbon (acetate, 300?mg/L) impaired the elimination of monochlorobenzene (removal of 72-96%). Microbial reductive dechlorination of perchloroethene was not detected in the rhizosphere under low carbon load, sulphate reduction, and high-carbon load conditions. Nonetheless, considerable amounts of perchloroethene were eliminated (79-87%), presumably by plant uptake and phytovolatilisation. Under fluctuating moderate carbon load conditions, perchloroethene dechlorination was initiated, and trichloroethene and cis-dichloroethene production showed that a minimum of 10% of the perchloroethene inflow load was dechlorinated. Sulphate reduction and the associated sulphide toxicity showed to constitute a hazard for constructed wetland treatment of sulphate containing groundwater contaminated with chlorinated volatile organic compounds, causing a decrease in removal efficiencies by 50 and 20% for monochlorobenzene and perchloroethene, respectively.