One of the primary responsibilities of governments is to shield citizens from life threatening events such as natural disasters. Previous work has demonstrated that voters electorally reward incumbents for the provision of disaster relief but not for investing in disaster prevention, leading incumbents to over-invest in relief spending even though investing in preparedness would yield large efficiency gains. This suggests that natural disasters offer governments a unique opportunity to improve their re-election prospects by providing generous disaster relief. Building on these findings, the present thesis first theorizes about and empirically explores which types of electorally motivated strategies guide the allocation of disaster relief. Using county-level data on natural disasters and relief payments in the context of U.S. presidential elections (1985-2008) the thesis shows that presidents channel relief aid to high-turnout counties in which they enjoy strong electoral support even when controlling for disaster damage. A subsequent analysis then investigates how citizens prefer policy benefits to be distributed among individuals experiencing natural disasters and to what extent presidents' distributive choices align with citizens' preferences. An exploration of data from a novel survey experiment with a representative sample of U.S. citizens suggests that citizens prefer allocations that reflect affectedness and need, but not electoral ties. Despite a notable degree of congruence between preferred and observed spending decisions, citizens' preferences do conflict with presidents' electoral considerations. Finally, the third part studies individual disaster policy preferences and investigates whether providing additional information about the effectiveness and efficiency of preventive measures affects support for preparedness spending. Two novel survey experiments with a representative sample of U.S. citizens indicate that i.