Floods are among the most damaging natural hazards and their impacts have been dramatically increasing worldwide over the past decades. As most basins of the world are ungauged or poorly gauged and some measurement networks are continuously under decline, the spatial distribution of flood hazard is often difficult to estimate because the input data needed for flood inundation modelling (e.g. topographies, flood extents, water levels) are often not available. A unique opportunity is nowadays provided by the ongoing development of remote sensing data, such as the low-cost, space-borne data. In particular, the development of new remotely sensed data sources has not only shifted flood modelling from a datapoor to a data-rich environment, but also provided a paradigm shift in flood modelling: from developing more sophisticated flood models to evaluating potential of remote sensing data. There is a general consensus that the increased availability and quality of those low-cost remote sensing data will be valuable for improving prediction in ungauged basins. However, their value and potential in supporting hydraulic modelling of floods are still not sufficiently explored in view of the unavoidable, intrinsic uncertainty affecting any modeling exercise. In this context, this thesis aims to explore the potential and limitations of low-cost, space-borne data in flood inundation modelling under uncertainty. In our research work, we analyze the potential in supporting hydraulic modelling of floods of: NASA’s SRTM (Shuttle Radar Topographic Mission) topographic data, SAR (Synthetic Aperture Radar) satellite imagery and radar altimetry. The characteristics of those data, and their pros and cons for inundation modelling are discussed. For example, SRTM`s global coverage and relatively low vertical error on low-slope areas are in favour of floodplain modelling, while its absence of in-channel geometry information would hamper its application in flood studies. Low-cost SAR imagery`s day-night, all-weather, cloud-free acquisition are particularly useful for flood extent monitoring, while its low resolution could induce equifinality in inundation model conditioning. Radar altimetry`s reliable water level measurements over large rivers provides opportunities for flood model calibration and evaluation, while its low space-time frequency limits the application in areas such as flood forecasting. To this end, research work has been carried out by either following a model calibration-evaluation approaches or by explicitly considers major sources of uncertainty within a Monte Carlo framework. To generalize our findings, three river reaches with various scales (from medium to large) and topographic characteristics (e.g. valley-filling, two-level embankments, large and flat floodplain) are used as test sites. Thus, specific modelling exercises are implemented with slight, tailor-made modifications to deal with practical issues, such as the actual data availability, the characteristics of flood events etc. The usefulness of the low-cost space-borne data is quantitatively analyzed. Lastly, an application of SRTM-based flood modelling of a large river is conducted to highlight the challenges of predictions in ungauged basins. The outcomes of the study provide indications on the potential and limitations of low-cost, space-borne data in supporting flood inundation modelling under uncertainty. Specifically, DEM resolution is often less of an issue than its vertical accuracy, as long as the coarse resolution allows the representation of flood patterncontrolling topographic features for the flood modelling issue, which is often not the case in urban flood studies. Thus, the thesis includes and discusses the usefulness of these data according to specific modelling purpose (e.g. re-insurance, planning, design). Moreover, topographic uncertainty could be compensated by other sources of uncertainties in hydraulic modelling if they are explicitly taken into account. The model prediction based on SRTM can be very close to that based on high-resolution, high-accuracy topographic data under other sources of uncertainty. However, besides modelling purpose and uncertainty considered, their actual usefulness could be affected by several other factors, such as the scale of the river under study, flood frequency, and the choice of modelling tools. Furthermore, the issue of in-channel information absent in SAR-derived DEMs are also discussed. It could be partially resolved by using either the global river depth dataset, or depth estimating from hydraulic geometry theory or model parameterization. Lastly, we discuss the upcoming satellite missions, which could potentially impact the way we model flood inundation patters.