This paper applies smooth transition regressions to incorporate nonlinearity into the impact of trading volume on exchange rate volatility, the so-called mixture distribution hypothesis (MDH). Linking this analysis to the Tobin tax debate, we provide the first empirical corroboration that such a tax may be effective in limiting speculation and reducing exchange rate volatility, especially in turbulent times. Our study points to two main results. First, we show that nonlinearities should be taken into account to explain the MDH. When volatility, spreads, and volume are simultaneously high, the relationship between trading volume and volatility tends to grow stronger and thus the MDH holds in turbulent periods. Second, on the assumption of constant trading volume elasticity, a Tobin tax would have been stabilizing and effective in the 2008 crisis.