Does partisanship matter for income inequality? The empirical evidence is notably mixed. I argue that these inconsistent findings are partly the result of scholars not (properly) modeling the relationship between unionization and the partisan composition of government. If we accept that union members favor greater government intervention to reduce economic inequalities than non-union members, politicians desire to hold elected office and be popular but also to implement particular public policies, and left party politicians share union members’ policy preferences to a greater extent than center and right party politicians, then partisan differences should be greatest at moderate levels of unionization, a condition that allows all politicians to pursue their policy preferences while maintaining electoral viability. The empirical analysis examines 16 wealthy democracies between 1970 and 2010. The results of error-correction models confirm the theoretical expectation and hold regardless of whether there is a majoritarian or proportional representation electoral system.