Abstract: We develop a theory of institutional transition from dictatorship to minority dominant-based regimes. We depart from the standard political transition framework à la Acemoglu-Robinson in four essential ways: (i) population is heterogeneous, there is a minority/majority split, heterogeneity being generic, simply reflecting subgroup size; (ii) there is no median voter in the post-dictatorship period, political and economic competition is favourable to the minority (fiscal particularism), reflecting group membership-based exclusion; (iii) (windfall) resources are introduced; (iv) we distinguish between labour income and resource windfalls, and labour supply is endogenous. We examine transitions via popular uprisings. Our theory predicts that if population polarisation is strong enough, the autocracy falls down irrespective of the level of resource windfalls. The latter only matter for an homogeneous enough economy: in such a case, the larger the resource windfalls, the lower the probability for the autocracy to fall. Finally, the theory highlights the role of labour market repression in authoritarian regimes as a strong engine of institutional change.