Poverty and inequality are primarily symptoms of exclusion – exclusion from employment, from education and from the economy. The twin drivers of transformation, economic growth and redistribution, should be effecting a reduction in poverty and narrowing inequality. However, more South Africans have sunk into poverty between 1996 and 2001, at least according to census data. Because there has been economic growth per capita, though, the combined effect is an unambiguous increase in inequality. The main article investigates shifts in long-term poverty and inequality trends as represented by a careful comparison of new data from the 2001 census with the 1996 census. Income-related changes in this period show a clear increase in the proportion of the population in poverty, though not in absolute poverty. Not only does a higher proportion of the population fall into this ‘middle’ group of the poor in 2001 compared to 1996, but on average people within the group are closer to the level of destitution. This is particularly true for the African and coloured population groups. Coupled with an increased proportion of Africans in the upper income groups, this means that the long-term trend of increasing inequality within population groups has accelerated. These findings need to be balanced against massive increases in access to services by the poor, particularly the poorest of the poor – although, as the data show, lack of income often prevents full take-up of the improved access. Such service increases were unequally balanced between provinces, with Limpopo residents enjoying faster increases than, for example, residents of the Eastern Cape. However, our understanding of poverty and the progress against it is only as good as our data. Steven Friedman points out in his commentary that data problems should not stall the urgency of action, but that improved and more regular data are critical to any assessment of the success of redistributive interventions. In a brief review of trend indicators since 2001, we argue that the increased take-up of grants alone, since the 1996–2001 period, together with accelerating economic growth, may be slowing further increases in poverty.