Extreme rainfall events pose a constant threat to society. There is evidence that the frequency of extreme rainfall events might increase as a consequence of climate change. Hence, it is important to analyze trends in extreme rainfall. Regional frequency analysis uses records from multiple sites to derive more precise estimates than could be obtained by at-site estimation. In this talk I will present a newly developed regional peaks-over-threshold model for non-stationary precipitation data. Furthermore, I will present a regional approach for threshold selection.