Abstract This article examines the effect of fiscal slack on government corruption using the US states in the period from 1998 to 2012 as a research sample. Fiscal slack in the US states is commonly referred to as “rainy day funds” (RDFs), which are intended as countercyclical reserve funds for government‐wide purposes. Theoretically, bureaucracy models predict that fiscal slack might catalyze the embezzlement or misuse behaviors of bureaucrats, who are considered to be budget maximizers. However, formally established and rules‐bound RDFs may function as a “strongbox” that curbs officials’ discretionary power, reduces uncertainty in fiscal slack management, and ultimately restrains embezzlement and misuse behaviors. Empirically, we use the incidences and durations of natural hazards as instrumental variables for RDF balances to address the potential endogeneity problems. We find that state RDFs help reduce government corruption, especially when they are regulated by relatively looser deposit rules and stricter withdrawal rules. Evidence for Practice State RDFs function as a “strongbox” for governments’ unspent budgetary resources and negatively affect the degree of government corruption. A one‐standard‐deviation increase in per capita RDF balances reduces corruption‐related convicted officials per million population by approximately 3 or, measured in an alternative way, reduces corruption‐related convicted officials per thousand government employees by approximately 44–50. The efficacy of state RDFs in curbing government corruption is contingent on the associated rules of operation. Specifically, RDFs bound by relatively looser deposit rules and stricter withdrawal rules reduce government corruption to a greater extent.