A numerical technique is developed for estimating water quality violation frequencies due to pollutant discharges from urban areas during combined sewer overflow events. The first four moments of in-stream pollutant concentration are found by integrating a pollutant loading - water quality impact model over the probability density functions of its stochastic rainfall and streamflow inputs. The Pearson family of probability distributions is then used to estimate violation frequencies of water quality criteria. The approach proves to be a flexible and efficient method for the initial screening of alternative combined sewer overflow abatement measures.