Demand for reducing mental health impacts from flooding through collective flood defence is elicited using a contingent valuation method with a sequential hypothetical scenario, which accounts for human resilience and experience. A two-step model fits the survey data: it combines a binary sample selection rule to distinguish protesters and participants with a Tobit model to accommodate true zero responses among participants. Results show that non-symptoms-specific information on mental health risk may bias the willingness to pay downward. Risk-averse individuals who have taken self-insurance protection measures are willing to pay for additional protection through collective defence. Feelings, such as worries and anxiety related to flooding, drive the demand, which supports the risk-as-feelings hypothesis for mental health protection from flooding. Inexperience rather than experience of flooding is found to increase demand, which indicates that individual mental resilience to flooding may increase after an event as posited by the inoculation hypothesis.