In this paper we assess the degree to which the current social security system redistributes income from rich to poor. We then estimate the impact of various proposed changes to social security on the overall redistributive effect of the system. Our analysis takes a steady-state approach in which we assume participants work their entire lives and retire under a given system. Redistribution is measured on a lifetime basis using estimated earnings profiles for a sample of people taken from the PSID. We allow for differential mortality, not only by gender and race, but also by lifetime income. Our results indicate that the current social security system redistributes less than is generally perceived, mainly because people with higher lifetime income live longer and therefore draw benefits longer. Remaining progressivity is reduced and even reversed by an increase in the assumed discount rate, since regressive taxes become more important relative to later progressive benefits. We find that many of the proposed changes to social security have surprisingly little effect on the redistribution inherent in the system.