The recent growth of Pentecostal Protestantism in Latin America has called attention to the social consequences of religious conversion. Ethnographic studies find that affiliation with Pentecostal churches is associated with attitudinal and behavioral transformations that modify gender relations, enhance the economic viability of the household, and endow families with social capital that can be mobilized during periods of economic stress, sickness, and emergency. The nature of the changes suggests that they may improve the welfare of infants and young children within the family. To explore this proposition, we use sample data from the 2000 demographic census in the Brazilian northeast to estimate the probability of death among children born to women 20 to 34 years of age. The findings show that, other things being equal, the death rate among children born to Protestant women is around 10 percent lower compared to the death rate among children born to Catholic women. We further disaggregate Protestants into "traditional" (e.g., Baptist, Presbyterian) and "Pentecostal" (e.g., Assembly of God) subgroups, and show that the mortality-reducing effect is greater among historical compared to Pentecostal Protestants. No mortality effects were associated with membership in neo-Pentecostal churches.Recent decades have witnessed an unprecedented increase in the number of people in Latin America who have converted from Catholicism to Pentecostal Protestantism. The emerging religious landscape is associated with changes in attitudes and behaviors that alter the relationship between men and women and modify intra-household behavior in ways that enable families to better cope with the varied injuries of poverty. The changes underway promote forms of behavior and social organization that are likely to improve the welfare of infants and children in the family.A commonly used indicator of child welfare is the probability of death in the early years of life. Because it responds to the joint effects of an array of social, economic, and biological phenomena (Birdsall 1980;Chen 1983), the child mortality rate is a sensitive indicator of child welfare. To test the hypothesis that the welfare of children is associated with religious affiliation we turn to sample data from the 2000 demographic census of Brazil, a predominantly Catholic country where the proportion of Protestants has risen dramatically in recent decades. We limit the analysis to the Brazilian northeast, the poorest region of the country where education is low, poverty is widespread, and child mortality is high. As we will show in the pages that follow, other things being equal, children born to Protestant mothers are significantly more likely to survive the risks of childhood compared to children born to Catholic mothers.Although census data enable us to estimate differences in child mortality according to mothers' religious affiliation, the information contained in the data does not permit us to precisely identify the reasons children born to Protestant mothers have a sur...