Integrated urban drainage modelling and environmental impact assessment require sewer emission models to be linked with submodels for treatment infrastructure and receiving rivers. The uncertainty in current water quality modelling is, however, huge, and environmental impact assessment looses more and more credibility. Based on an integrated modelling case for a combined sewer \342\200\223 WWTP \342\200\223 river system, it is shown in the paper that the integrated model does not produce more accurate results in comparison with the random simulation of emission concentrations from a frequency distribution. This should, however, not pose a serious problem as in most applications of impact assessment model results are not needed in real time but in statistical terms. Further investigation makes clear that detail/sophistication in water quality modelling is not so important, but that more focus has to be given to long-term simulations, the use of parsimonious models and model validation based on concentration frequencies.