Socioeconomic status (SES) and health during childhood have been consistently observed to be associated with health in old age in many studies. However, the exact mechanisms behind these two associations have not yet been fully understood. The key challenge is to understand how childhood SES and health are associated. Furthermore, data on childhood factors and life course mediators are sometimes unavailable, limiting potential analyses. Using SHARELIFE data (N = 17230) we measure childhood SES and health circumstances, and examine their associations with old age health and their possible pathways via education, adult SES, behavioural risks, and labour market deprivation. We employ structural equation modelling to examine the mechanism of the long lasting impact of childhood SES and health on later life health, and how mediators partly contribute to these associations. The results show that childhood SES is substantially associated with old age health, albeit almost fully mediated by education and adult SES. Childhood health and behavioural risks have a strong effect on old age health, but they do not mediate the association between childhood SES and old age health. Childhood health in contrast retains a strong association with old age health after taking adulthood characteristics into account. This paper discusses the notion of the ‘long arm of childhood’ and concludes that it is a lengthy, mediated, incremental progression rather than a direct effect. Policies should certainly focus on childhood, especially when it comes to addressing childhood health conditions, but our results suggest other important entry points for improving old age health when it comes to socioeconomic determinants.