This study present the first full scale application of an integrated Model Predictive Control (MPC) strategy for optimization of urban drainage networks. The strategy was implemented in a 76 km2 urban catchment in the centre of Copenhagen (Denmark). This generalized strategy combines runoff forecast generated by radar-based models with real-time water level measurement across the drainage system. A dynamic overflow risk assessment is performed by taking into account the estimated level of uncertainty of runoff predictions. The cost associated to the expected overflow events is minimized by a genetic algorithm. The MPC system is implemented in a commercially available platform, enabling the future connection between the dynamic strategies controlling urban drainage systems and wastewater treatment plants. Full-scale implementation is almost completed and the preliminary results from off-line test show the potential for application of this methodology in a wide range of different urban areas.