The governments of developing countries have been decentralizing. Since the mid-1980s, decentralization reforms have been introduced in states ruled by virtually all varieties of national political regimes, from military dictatorships, authoritarian presidencies, and monarchies through single-party or dominant-party regimes to multiparty competitive democracies. There is no close connection between form of decentralization and regime type: elected, democratic local governments have been introduced by military regimes-mostly nonpartisan but even in one case (Bangladesh in the 1980s) with party competition-while technocratic, administrative deconcentration is to be found in a formally multiparty system such as that of Kenya, and, under Ghana's multiparty regime, which produced a peaceful transfer of power nationally in 2000, parties are banned from District Assembly elections. Much of the literature tends to assume that decentralization of government has generic benefits that are independent of regime context. The most common argument is that, because decentralization by definition involves bringing government closer to the governed in both the spatial and institutional senses, government will be more knowledgeable about and hence more responsive to the needs of the people.1 It is the more recent tendency to conflate decentralization with democratization and enhancement of participation at the "community" level, which underlies the belief that decentralization will lead to greater responsiveness to the needs of the "poor." Insofar as the majority of the population in developing countries is both poor and excluded from national elite or "high" politics, then any scheme that appears to offer greater political participation to ordinary citizens seems likely to increase their "voice" and hence (it is hoped) the relevance and effectiveness of government policy. In this chapter, we address two linked questions: whether decentralized forms of government in general are more responsive to poor people, and whether there is any systematic relationship between variations in responsiveness and the political and regime context of decentralized systems. It will be argued that regimes, particularly the kinds of policy coalitions constructed within democratic ruling parties at both the central and local levels, do have an important impact on the pro-poor outcomes of decentralization. We shall focus primarily on political and administrative decentralization, that is, the allocation of power among territorially defined and nested hierarchies.2 The majority of cases examined will be either devolved local governments with or without federal systems or mixed forms of devolution with deconcentrated administrations. Pure deconcentrated administrations, whether general or sectoral, are of less interest unless (as is increasingly routine) they incorporate at least some participatory elements (or aspirations). A review of the literature led to the selection of the following cases for which there was sufficient evidence to make a comparison of their responsiveness and social and economic performance outcomes: West Bengal and Karnataka States (India), Colombia, the Philippines, Ceara State (Brazil), Chile, Côte d'Ivoire, Bangladesh, Ghana, Kenya, Nigeria, and Mexico. The Indian states of West Bengal and Karnataka stand out as probably the most systematically studied and richly documented of any in the world, while the Philippines is beginning to generate a mass of research evidence that has yet to come to fruition. There are other interesting cases from the rest of Asia, Africa, and Latin America, but the evidence is frequently partial and contradictory. Bangladesh, Ghana, Côte d'Ivoire, Brazil, and Colombia count as among the best-documented cases from those continents, while Chile, Kenya, Nigeria, and Mexico provide useful contrasts on particular dimensions.