The objective of this paper is to understand the determinants of the enforcement level of indirect taxation in a positive setting. We build a sequential game where individuals, who di¤er in their willingness to pay for a taxed good, vote over the enforcement level. Firms then compete à la Cournot and choose the fraction of sales taxes to evade. We assume in most of the paper that the tax rate is set exogenously. Voters face the following trade-o¤: more enforcement increases tax collection but also increases the consumer price of the goods sold in an imperfectly competitive market.We obtain that the equilibrium enforcement level is the one most-preferred by the individual with the median willingness to pay, that it is not a¤ected by the structure of the market (number of …rms) and the …rms' marginal cost, and that it decreases with the resource cost of evasion and with the tax rate. We also compare the enforcement level chosen by majority voting with the utilitarian level. In the last section, we endogenize the tax rate by assuming that individuals vote simultaneously over tax rate and enforcement level. We prove the existence of a Condorcet winner and show that it entails full enforcement (i.e., no tax evasion at equilibrium). The existence of markets with less than full enforcement then depends crucially on the fact that tax rates are not tailored to each market individually.