The design of UV/H2O2 reactors is studied. First an analytical approach is followed that provides characteristics (maximum fluence and maximum possible degradation) for an ideal system. An efficiency parameter is introduced that relates the actual performance of a reactor with the best possible performance for a reactor with a certain water flow, lamp power and UV transmittance. From the analytical model, several design parameters were investigated. The desired treatment level influences the choice of a design parameter, as the fluence distribution becomes less important at lower desired treatment levels. As the desired degradation levels in UV/H2O2 applications are lower than the desired inactivation levels for UV disinfection, a UV/H2O2 reactor may be designed with a larger water layer depth than UV disinfection reactors. Also, manipulating the velocity profile towards a profile that mimics the fluence rate profile is beneficial, as well as increasing mixing. Increasing the number of lamps (while the total energy consumption remains the same) is beneficial, as it results in a more uniform fluence rate profile. Enlarging the quartz sleeve has limited effect. Changing the distribution of hydrogen peroxide in the reactor also has limited effects. Using the design parameters, new UV/H2O2 reactor types were developed with CFD simulations and tested experimentally. An increase in log degradation up to 30% was demonstrated by the improved reactor design. (C) 2015 Elsevier Ltd. All rights reserved.