This paper compares a standard expenditure-based poverty measure with a spe
cifically created composite measure of deprivation using household sun;ey d
ata from South Africa. While there is a strong overall correlation between
expenditures and levels of deprivation, the correlation is much weaker amon
g the worst-off South Africans. In addition. the two measures differ consid
erably in the impact of race, headship, location turban, rural), and househ
old size on expenditure poverty versus deprivation. In general, the depriva
tion measure finds more Africans, rural dwellers, members of de fitc ro fem
ale-headed households. and members of smaller households deprived than expe
nditure poor. Only the differences in the effect of household size on pover
ty are sensitive to assumptions about equivalence scales. As a result, the
two measures diverge greatly in identifying the poorest and most deprived s
ections of the population, which may have considerable consequences for tar
geting.