Inequality is generally thought to affect the electoral fortunes of the left, yet the theory and evidence on the question are unclear. This is the case even in Latin America, a region marked by enormous inequalities and by the stunning return of the left over the last decade. We address this shortcoming. Our game-theoretic model reveals that the probability that the left candidate is elected follows an inverted U-shaped relationship. At low levels of inequality, the rich do not bribe any voters and poor voters are increasingly likely to vote for the left candidate based on redistributive concerns. At high levels of inequality, the rich want to avoid redistribution and bribe poor voters, causing the left candidate to be elected with decreasing probability. We find support for our hypothesis, using 110 elections in 18 Latin American countries from 1978 to 2008.