A central pillar of contemporary political science is the assumption that poorer voters, in a democratic setting and when given the opportunity to do so, will support punitive taxes on the wealthy to finance downwards redistribution (Meltzer and Richard 1981). This interpretation of democracy as a modern day 'Robin Hood' has proven crucial for our understanding of when, and how, transitions to democracy occur (Ansell and Samuels 2014). However, we argue that the relationship between inequality and redistributive preferences has been misunderstood. In countries where inequality is endemic, citizens will not expect redistribution to result from taxation. And why steal from the rich to help no one? We expect that citizens in unequal countries will view taxation primarily as a burden and will not express much appetite to use it as a tool for redistribution. Moreover, aversion to `soaking the rich’ will be especially high in contexts where people already feel over-taxed. We test these expectations first through conjoint experiments in Mexico City and Bogotá, Colombia. This controlled comparison allows us to compare two highly unequal contexts that share similar economic and political profiles but diverge in their levels of income tax. We then test the generalizability of our claims through a global analysis using the World Values Survey, and we find increasing taxes in highly unequal countries has a negative effect on support for redistributive taxation.