Polarization -- defined as the difference in policy preferences along the salient axis of political competition -- among non-regime elite actors has important consequences for successful democratic consolidation during authoritarian transitions. In contingent theories of democratization, higher levels of polarization are less likely to facilitate the elite cooperation and compromise crucial to initial transition periods, and empirical evidence from potential democratic transitions following the ‘Arab Spring’ and other uprisings in comparative cases support these theories. Yet existing studies leave undertheorized why elites emerge more or less polarized from an authoritarian regime. In this paper, I present an original theory of how the repression of formal political opposition that defines authoritarian regimes affects processes of polarization in these systems. The theory builds on social psychology findings about the causes and consequences of group identification to hypothesize that the nature of repression -- whether it targets a specific group or is more widespread -- alters individual opposition groups’ levels of in-group identification, in turn altering levels of polarization in political preferences among these groups. I test the theory through case studies of Egypt and Tunisia, relying on original evidence drawn from extensive interviews, political memoirs, party platforms, and internal party documents. First, I detail how repression differed under successive authoritarian regimes in Egypt (a targeted repressive environment) and Tunisia (a widespread repressive environment). I then analyze how the different repressive environments altered both the nature and level of group identification within/among opposition groups over time, and resulted in different levels of political polarization within each political system. I also show how understanding polarization as the product of authoritarian repressive legacies has significant explanatory value for why elite actors converged on similar policy preferences and compromised in Tunisia, but failed to do so in Egypt following 2011 “Arab Spring” uprisings. The paper advances a different approach to elite polarization, focusing on the political psychology of elites rather than employing a rational actor model to explain preferences and behaviors. In addition, it rethinks dominant regime-repression models, in considering repression’s long term effects after regime transitions and on the identities, rather than the tactics, of those groups targeted. Finally, my findings challenge current understandings of authoritarian politics in the Middle East as well as comparative transitology theories, in considering the psychological and political legacies of authoritarian repression on subsequent political developments.