The evidence shows that government spending for health in many developing countries benefits the well-to-do more than the poor. However, a combination of favorable political forces and sound public policies can shift the focus of government expenditures toward the poor. Doing this is an essential part of any effective poverty reduction program in developing countries. The better-off gain disproportionately in part because most public spending goes toward curative, high-cost hospital care in urban areas, where income is usually above average and where the wealthy have better acces to care. In addition, health care subsidies for the middle classes, in the form of tax relief and government transfers to social security-based health insurance, skew total public expenditure further toward the better-off. This inequity can be understood in the context of "public choice" theory: politicians and bureaucrats seek not to optimize economic efficiency, but instead to maximize their chances of getting reelected or staying employed. This paper suggests that to implement health policies that favor the poor, it is necessary to "swim against the tide." Costa Rica, Korea, Malaysia and Zimbabwe provide examples of success stories with lessons for other countries. High and sustained economic growth has set the stage for redistributive health spending. So have pro-poor political forces. In Zimbabwe, for example, these forces were represented in post-independence groups eager to redress earlier discrimination. In Korea, it was a modernizing authoritarian regime in search of greater legitimacy. In political environments like these, pro-equity governments have reduced, eliminated or avoided health care subsidies that blatantly or covertly favored the better-off. Public spending has been targeted toward initiatives serving primarily the poor, such as rural health facilities and basic primary and preventive care. In middle-income countries like Costa Rica, Korea and, to a certain extent, Chile, increased public spending has universalized health insurance. In these circumstances, general tax revenues have come to subsidize insurance for the poor more than for the middle class. The authors argue that donnor involvement can help to implement policies and programs to improve equity in health. They also advocate expanded donor backing for applied research to measure exisiting inequalities and to analyze the political dynamics behind the distribution of public resources for health.