Social capital has been identified as an important avenue of upward mobility for poorer people. However, recent theoretical work suggests that in highly polarized societies, the accumulation of social capital is likely to be fragmented and ineffective for people at the bottom of the economic pyramid. In South Africa, apartheid-era policies created such deep, socially embedded inequality producing a self-reinforcing circle of social exclusion and persistent poverty as another of apartheid's legacies. Work to date on post-apartheid income distribution-with its demonstration of increasing inequality and poverty-is consistent with this legacy hypothesis. This paper takes this hypothesis further by using a two-pronged approach that draws on quantitative and qualitative data to explore the role of different types of assets in explaining poverty status. First, novel econometric analysis of poverty and livelihood dynamics is used to test for a poverty trap that would signal the existence of a ceiling to upward mobility for poor people. The analysis finds evidence of such a trap. Secondly, the qualitative data is used to confirm and more deeply probe the reasons behind the patterns of truncated upward mobility, finding accessibility and stability of employment and state pensions as key factors explaining why people remain poor or non-poor. While this analysis finds ample evidence of active social capital and networks, these are more helpful for non-poor households while for the poor they seem to at best help stabilize livelihood at low levels and seem to do little to promote upward mobility. This paper's confirmation of the legacy hypothesis suggests the publicly provided social safety nets that exist in South Africa need to be at least maintained if not strengthened, while state policy needs to take a more aggressive role in assuring that households have access to a minimum bundle of assets and to the markets needed to effectively build on those assets over time.