Substantial cross-national differences in poverty alleviation are well documented, but theextent to which different parts of the social transfer system account for this variation is still relatively unexamined. This study analyses the redistributive effects of specific social policy institutions in a comparative perspective. The main question is to what extent non-targeted provisions and means-tested benefits reduce relative economic poverty in different institutional settings. It is shown that the structure of non -targeted entitlements is more important than that of means-tested benefits in explaining differences in poverty alleviation across countries. The study also presents a new method for estimating the anti-poverty effects of separate parts of the social transfer system. This method decomposes the anti-poverty effects of a set of social transfers into independent and combined effects, which produces more valid results than prevalent methods used to assess the impact of a particular transfer on poverty. The countries included in this study are Canada, Germany, Sweden, the United Kingdom and the United States. The empirical analyses are based on the Social Citizenship Indicators Program (SCIP) and Luxembourg Income Study (LIS) for data points describing the situation in the mid-1990s.