We use event history analysis to separately model a state’s first and then repeated tax amnesties and test the revenue yield versus the fiscal stress hypotheses using a panel of annual data from all 50 states for the period 1982–2010. We also split our sample into two sub-periods, 1982–1988 and 1989–2010, to more easily compare our results to other studies and to see if there is a significant difference between early tax amnesties and more recent ones. We find that state fiscal stress is more important than the potential yield from an amnesty, particularly for more recent amnesties. Our findings contrast with the existing literature which found evidence that states are likely to enact tax amnesties due to a revenue yield motive rather than a fiscal stress motive.