South Africa is an example of a developing country where there is substantial variation in household size and composition by income class and by race. However, inequality measures which are typically derived from per capita adjustments are insensitive to these variations. In this study, we estimate equivalence scales for South Africa using Engel’s method applied to expenditure micro-data. Our primary objective is to show how the use of the estimated scales has sizeable effects, significantly reducing existing estimates of total inequality as well as both inter- and intra-racial inequality. For example, the Gini coefficient, measuring total inequality, falls by six percentage points from 0.63 to 0.57 in 2008/2009 when we use the estimated equivalence scales. These findings have important implications for the measurement and comparison of inequality measures over time.