Older people now constitute the majority of those with health problems in developed countries so an understanding of health variations in later life is increasingly important. In this paper, we use data from three rounds of the Health Survey for England, a large nationally representative sample, to analyse variations in the health of adults aged 65-84 by indicators of attributes acquired in childhood and young adulthood, termed personal capital; and by current social resources and current socio-economic circumstances, while controlling for smoking behaviour and age. We used six indicators of health status in the analysis, four based on self-reports and two based on nurse collected data, which we hypothesised would identify different dimensions of health. Results showed that socio-economic indicators, particularly receipt of income support (a marker of poverty) were most consistently associated with raised odds of poor health outcomes. Associations between marital status and health were in some cases not in the expected direction. This may reflect bias arising from exclusion of the institutional population (although among those under 85 the proportion in institutions is very low) but merits further investigation, especially as the marital status composition of the older population is changing. Analysis of deviance showed that social resources (marital status and social support) had the greatest effect on the indicator of psychological health (GHQ) and also contributed significantly to variation in self-rated health, but among women not to variation in taking three or more medicines and among men not to self-reported long-standing illnesses. Smoking, in contrast, was much more strongly associated with these indicators than with self-rated health. These results are consistent with the view that self-rated health may provide a holistic indicator of health in the sense of well-being, whereas measures such as taking prescribed medications may be more indicative of specific morbidities. The results emphasise again the need to consider both socio-economic and socio-psychological influences on later life health.