Conventional wisdom suggests that global governance through international organizations (IOs) is growing increasingly polarized—while democracies group together in some IOs, autocracies gather in others. Yet, so far, research offers little insight into the extent of this phenomenon and the factors that might be driving it. This paper offers the first systematic comparative analysis of regime-based sorting into IOs. We conceptualize sorting as situations when states opt for membership in IOs whose members share the same regime type. Theoretically, we develop expectations about why and under what conditions states engage in sorting. Empirically, we test these expectations on all formal and informal IOs for the time period 1925–2010. Our findings have implications for contemporary debates on democratic backsliding, international cooperation, and challenges to global governance.