Flood risk management is the process of analysing, assessing and (if required) reducing flood risks, in terms of economic costs and affected population or loss of life. The analysis is performed through a probabilistic approach in which the factors are represented by a probability distribution function to address uncertainties. However, relevant factors in the analysis, such as mean sea level and the population living in the area, often are deeply uncertain on the longer term, resulting for example from climate change and changing socio-economic conditions. Generally, FRM explores these conditions through scenario analysis. However, the way FRM creates and analyses the scenarios is subjective and leaves open a lot of uncertainty. Often, only a limited set of subjective qualitative scenarios is made, which risks not taking scenarios into account that turn out to be important later. Quantitative scenarios become increasingly uncertain as the time horizon over which these are used lengthens, due to uncertainty in the trend.This research investigates a new addition to flood risk management to support it in dealing with long-term deep uncertainty. It uses robust decision making approaches as an addition to the usual FRM approach, by analysing results for a much larger set of scenarios to increase robustness, both static and flexible, to different scenarios. It combines aspects of dynamic adaptive policy pathways (Haasnoot, Kwakkel, Walker, & ter Maat, 2013) and aspects from robust decision making (Lempert, Popper, & Bankes, 2003), with the FRM approach to create an eight-step iterative approach called Flood Risk Adaptation Pathways, which was later tested in a case study. The approach starts by analysing the current situation and possible futures, after which it determines the difference between those situations and the preferred state. Based on this, possible flood risk reduction measures are designed, and then tested on their performance in a flood risk screening model, to create so called effectiveness regions, which show under which conditions which solutions are effective in reducing the gap between states. Based on the effectiveness regions, pathways of actions over the different possible futures are created. A monitoring plan is then created, which defines the signposts and their trigger values. Finally, the situation is monitored and actions are implemented when required.The Flood Risk Adaptation Pathways approach aims to provide policymakers with the ability to anticipate different futures, by creating actions, which meet the objectives, for a wide range of tested futures. It should allow policymakers to plan the implementation of measures ahead of time. The long-term situation is investigated, including the measures required at that point, and linked to the current situation, including the currently required measures. This should ensure that measures remain implementable by actively assuring the conditions for implementation are not violated, thus reducing negative effects of path-dependency. The link between the long-term measures and the current measures should also allow for an improved cohesion of the sequence of measures, with short-term measures providing a starting point for long-term measures. Finally, it narrows down the measures to the ones highlighted by the approach for later on in the process, when an action needs to be implemented.The approach has been tested in a case study, on Beira, Mozambique, which is prone to severe rainfall and coastal flooding. Furthermore, compound flooding problems arise when high sea water levels coincide with rainfall events. The Flood Risk Reduction Evaluation and Screening model (van Berchum, van Ledden, Jonkman, Timmermans, & van den Broek, 2019) was used to model the current flood risk, as well as the effect of flood risk reduction measures on the flood risk, affected population and construction costs. For this research, the model was adapted to test specific actions over a set of changing conditions, to test their performance for the longer term. The approach was then used in the case study, following the steps (described above). The data was gathered from local sources and reference projects. After the completion of the case study, the approach was reflected upon: The main limitation arising from the case study, is the considerable computation time and time-consuming analysis of the results. Overall, Flood Risk Adaptation Pathways strengthen the FRM approach in dealing with long-term deep uncertainty, and thereby enhance the performance of flood risk reduction strategies under different than expected conditions.