This article revisits the oft-cited relationship between economic shocks and coups. According to conventional wisdom, economic recessions trigger coups. However, existing empirical studies have not consistently produced supporting evidence for that relationship. This article claims that this is partly because existing studies have not differentiated transitory from permanent shocks to the economy. Two different economic shocks could have different effects on coups. Moreover, existing studies have not sufficiently addressed measurement error in gross domestic product (GDP) data. To overcome these problems, I use exogenous rainfall and temperature variation to instrument for economic growth. Instrumental estimates demonstrate, consistently across four different GDP per capita growth measures, that a decrease in GDP per capita growth rates, induced by short-run weather shocks, significantly increases the probability of a coup attempt. Conversely, noninstrumental variable estimates vary according to different GDP measures, and are close to zero, consistent with previous findings.