In the city of Berlin combined sewer overflows (CSO) can lead to severe depressions in dissolved oxygen (DO) of receiving urban rivers and hence to acute stress for the local fish fauna. To quantify CSO impacts and optimize sewer management strategies, a model-based planning instrument has been developed. It couples the urban drainage model InfoWorks CS which simulates hydraulics and pollutant transport in the sewer with the river water quality model QSim which simulates hydraulics, mass transport and various biogeochemical processes in the receiving water body. To identify simulated CSO impacts, concentration-durationfrequency-thresholds for DO are applied to river model results via an impact assessment tool. Two kinds of impacts are distinguished: i) suboptimal conditions and ii) critical conditions for which acute fish kills are possible. In the case of Berlin, suboptimal conditions are observed on up to 92 days per year, predominantly during periods of low discharge and high temperatures whereas critical conditions only occur after CSO. For model calibration and validation, continuous measurements in both river and sewer are used. First simulations show good accordance between simulated and measured DO concentration in the river with Nash-Sutcliffe efficiencies between 0.70 and 0.79 for an eight-month time period at three different river monitoring points. However, to assure satisfactory model performance for adverse DO conditions in particular, impact assessment results for measured and simulated data are compared. Regarding suboptimal DO conditions simulated and measured data show good agreement. Nevertheless model representation for critical conditions is poor for some river sections and requires further improvement for CSO conditions. The results underline the importance of combining different validation approaches when dealing with complex systems.