Although natural hazard insurance is advocated as an important means of risk management, private insurance demand often remains below critical levels. Prior loss experience and the design of governmental relief schemes are two factors potentially influencing insurance decisions. We address these two elements in monetary incentivized experiments which include representations of natural hazard insurance schemes in Europe. We draw on two very different samples: First, we run a laboratory experiment with a student subject pool in Germany. In addition, we replicate the experiment as an online experiment with citizens of flood-prone areas in the city of Dornbirn (Austria). The experiment reflects two possible designs of governmental relief schemes: partial but guaranteed relief and full but nonguaranteed relief. The risk of loss is kept constant over ten consecutive rounds to analyze the effect of loss experience. In both of our samples, the design of the governmental relief scheme has no effect on insurance decisions. Furthermore, prior loss experience adversely affects insurance decisions. Uninsured subjects tend to remain uninsured after experiencing a loss, and previously insured subjects often switch to non-insurance in the rounds after the loss. These results have important policy implications, e.g., for the optimal design of flood risk communication.