This article examines why democratic competition sometimes fails to curb governmental corruption. We argue that in democracies party system competitiveness, which shapes the ability of voters to effectively select and control their politicians through elections, plays a critical role in conditioning the scope for corruption. For voters, governmental corruption is a classical principal–agent problem and its magnitude is mediated by the extent to which the competitiveness of a party system helps to make information and effective choices available to the electorate. Informed voters who can coordinate on credible alternatives to under-performing and corrupt incumbents, we argue, can select politicians who are likely to curb corruption and hold accountable those who do not. We test this argument through a controlled comparative analysis of corruption in 70 democracies around the world and find broad support for our hypotheses.