Group polarization, or the progressive segregation of a population into two oppositional groups, appears to be happening in parts of the Muslim world; in some places Sunni and Shia are polarizing, and in others, Islamists and secularists. Group polarization is in fact evident in many times and places and (1) may spread transnationally and (2) may trigger foreign interventions. It is fruitful to understand group polarization as endogenous preference formation; i.e., the more social interaction takes place, the more individuals identify more strongly with their own type and against an opposing type; these identifications in turn affect individuals’ preferences over the allocation of goods, such that each progressively favors more goods for his own group and fewer for the opposing group. A complication arises from the fact that people have multiple group affiliations: exogenous events can cause a population to polarize along one identity axis (e.g., ethnicity) and de-polarize along another (e.g., nationality). Endogenous group polarization (EGP) contrasts with rational-choice accounts that see group polarization as the revelation or clarification of exogenous, fixed preferences. Two historical cases of transnational polarization – Europe 1917-22 and Africa’s Great Lakes region 1972-94 – can help adjudicate how helpful EGP and rational choice are, respectively, in explaining group polarization. In the end, an adequate model might need to recognize both that some actors (elites) have fixed preferences but that they are constrained by the endogenous preferences of other, non-elite actors.