A method for stochastic analysis of the effect of combined sewer overflows on the dissolved oxygen concentration in receiving rivers is developed. The method is based on repeated operation of the water quality model MOUSE-DOSMO. Each operation of this model covers a series of overflow events calculated by the MOUSE-SAMBA runoff model based on a historical rainfall record. For each event selected input data and model parameters in both models are drawn from fundamental statistical distributions by a simple Monte Carlo method. These data and parameters vary from event to event in each series. Each operation of the MOUSE-DOSMO model results in an extreme event statistics on dissolved oxygen minimum values -while repeated operation of the model yields extreme event percentiles by means of which a probability-based assessment of the model result may be performed.