A method is described for investigating life course influences on health in early old age. The lives of some 300 individuals at present aged 65-75 y have been reconstructed from the archived records of a pre-WWII survey, in which they took part as children, and from lifegrid interviews with the same individuals 60 y later. Despite loss to study at several points those interviewed are shown to be representative of the British population socio-demographically, in comparison with the 1931 and 1991 decennial censuses, and physically, in comparison with the Health Survey for England. Bias is conservative because the most disadvantaged were disproportionately affected by loss to follow-up through death and because non-responders to interview were more disadvantaged as children than the interviewees. Representativeness and conservative bias, it is argued, justify the use of these data for investigating life course influences on health in early old age.