Temporal and spatial variability in the rainfall field is regarded as a substantial source of uncertainty in flood prediction with spatially distributed rainfall-runoff models. But often -due to a lack of data - the influence of rainfall uncertainty on the accuracy of flood prediction is not properly addressed. In our case, we can base our uncertainty assessment on ensembles of 100 rainfall realisations for several extreme flood events, which were produced by stochastic rainfall simulation in hourly resolution. This data set allows us to investigate the specific sensitivity of the rainfall-runoff models NASIM, LARSIM, and WaSiM-ETH in a small catchment.