A number of studies have found that membership in a higher socioeconomic status (SES) group has a significant effect on different demographic outcomes such as lower infant and child mortality. Most of these studies have analyzed the association between SES and children's survival by focusing on data on asset-ownership which includes, for example, owning a bicycle, radio or television; housing characteristics such as number of rooms or type of toilet facilities; and source of water. These household characteristics are conceived as having a direct or indirect role in shaping child mortality differentials. This paper uses principal components analysis to create a living standards index (LSI) based on the household characteristics and apply it in a multivariate model to examine its relationship with childhood mortality in Malawi using 1987 and 1998 census data. When the LSI is applied to the 1987 census data, the results show an increase in mortality for children who come from poor households. However, the results in 1998 differ from those in 1987 in that child mortality is higher among the rich households in 1998 and also among middle-aged women. These results are consistent with parallel analysis of the 1992 and 2000 Malawi Demographic and Health Survey data. We argue that based on the magnitude of the HIV/AIDS prevalence in Malawi and given the timing of the 1998 census in the stage of the AIDS epidemic, and also consistent with findings of high mortality in all households and high social class groups in Malawi, the shift in the effect of the LSI on child mortality may be attributed to this deadly disease.