In the face of increased flood risk responsible authorities have set out safety margins to incorporate climate change impacts in building robust flood infrastructure. Using the case study of four catchments in Ireland, this study subjects such design allowances to a sensitivity analysis of the uncertainty inherent in estimates of future flood risk. Uncertainty in flood quantiles is quantified using regionalised climate scenarios derived from a large number of GCMs (17), forcedwith three SRES emissions scenarios. In terms of hydrological response uncertainty within and between hydrological models is assessed using the GLUE framework. Regionalisation is achieved using a change factor method to infer changes in the parameters of a weather generator using monthly output from the GCMs, while flood frequency analysis is conducted using themethod of probability weighted moments to fit the Generalised Extreme Value distribution to ~20,000 annual maximia series. Sensitivity results showthat for low frequency events, the risk of exceedence of design allowances is greater than for more frequent events, with considerable implications for critical infrastructure. Peak flows for the five return periods assessed were found to be less sensitive to temperature and subsequently to potential evaporation (PET) than to rainfall. The average width of the uncertainty range for changes in flood magnitude is greater for low frequency events than for high frequency events. In all catchments there is a progressive increase in the peak flows associated with the 5, 25, 50 and 100-year return periods when moving from the 2020s to the 2080s.