We show that tax-induced increases in alcohol prices can lead to substantial substitution and avoidance behavior that limits reductions in alcohol consumption. Causal estimates are derived from a natural experiment in Illinois where spirits and wine taxes were raised sharply and unexpectedly in 2009. Beer taxes were increased by only a trivial amount. We construct representative and consistent measures of alcohol prices and sales from scanner data collected for hundreds of products in several thousand stores across the US. Using several difference-in-differences models, we show that alcohol excise taxes are instantly over-shifted by a factor of up to 1.5. Consumers react by switching to less expensive products and increase purchases of low-tax alcoholic beverages, thus all but offsetting any moderate, tax-induced reductions in total ethanol consumption. Our study highlights the importance of tax-induced substitution, the implications of differential tax increases by beverage group and the impacts on public health of alternative types of tax hikes whose main aims are to increase revenue.