This paper presents the case for an ‘End-to-End’ flood inundation modeling strategy: the creation of a coupled system of models to allow continuous simulation methodology to be used to predict the magnitude and simulate the effects of high return period flood events. The framework brings together the best in current thinking on reduced complexity modeling to formulate an efficient, process-based methodology which meets the needs of today's flood mitigation strategies. The model chain is subject to stochasticity and parameter uncertainty, and integral methods to allow the propagation and quantification of uncertainty are essential in order to produce robust estimates of flood risk. Results from an experimental application are considered in terms of their implications for successful floodplain management, and compared against the deterministic methodology more commonly in use for flood risk assessment applications. The provenance of predictive uncertainty is also considered in order to identify those areas where future effort in terms of data collection or model refinement might best be directed in order to narrow prediction bounds and produce a more precise forecast.