I investigate whether cash thresholds that forbid cash payments on big transactions are effective at reducing tax evasion. I find that the 1000 euros threshold implemented in Italy in 2011 induced a bigger cash expenditure reduction for the households with self employed members, and the more so in case they work in cash intensive sectors. With the help of a simple model, I show that this empirical evidence suggests a tax evasion reduction, and I compute the tax revenue increase implied by the empirical estimates. Calibrating the model, I also perform a counterfactual exercise to quantify the potential effects of lower thresholds.