A common approach when studying inequalities in health is to use a wealth index based on household durable goods as a proxy for socio-economic status. We test this approach for elderly health using data from an aging survey in a rural area of South Africa and find much steeper gradients for health with consumption adjusted for household size than with the wealth index. These results highlight the importance of the measure of socioeconomic status used when measuring health gradients, and the need for direct measures of household consumption or income in ageing studies.