There is abundant evidence of a large association between many measures of socio-economic status, including income and education, and a variety of health measures later in life. Understanding the complex relationship between these socio-economic measures and health and mortality is critical for the development of strategies for improving population health – especially in an ageing society. However, the mechanisms and pathways by which differences in socio-economic status lead to different health outcomes later in life are largely unknown. These strong associations do not, however, necessarily imply causation. Indeed, the association between health and socio-economic status may be explained partially by selection on confounding factors, such as intelligence and parental background, which affect both the socio-economic status and health later in life. The main aim of this project is to estimate the causal effect of socio-economic status and its implied behavioral process on health and mortality later in life, accounting for selection issues both early and later in life.