This study contributes to the understanding of behavioural responses to climate change induced extreme weather events. It suggest a microeconometric method for measuring flooding related risk preferences of affected individuals. The method is outlined using the empirical case of agricultural production in floodplains of the UK over 28 years. A quasi-experimental approach to measure differences in the risk attitudes of farmers located in high flooding risk areas versus farmers located in low flooding risk areas is followed. Changes in flooding risk related behaviour over time is analysed and marginal effects of different individual and disaster related characteristics for this behaviour are investigated. Beside a moments based risk estimation approach the study also applies a dynamic panel estimator. The estimates suggest that the average farmer located in a high flooding risk area is prepared to pay about 6% more of his profit for insuring against the higher risk of flooding compared to farmers in low flooding risk areas. The significance of considering individual risk preferences for an efficient flood policy design is discussed using the example of voluntary agreements for the maintainance of flood defences.