Interest among historians, philosophers and sociologists of science in population-based biomedical research has focused on the randomised controlled trial to the detriment of the longitudinal study, the temporally extended, serial observation of individuals residing in the same community. This is perhaps because the longitudinal study is regarded as having played a secondary role in the debates about the validity of populations-based approaches that helped to establish epidemiology as one of the constitutive disciplines of contemporary biomedicine. Drawing on archival data and publications relating to the Baltimore Longitudinal Study of Aging, we argue however that the historical development of the longitudinal study is richer and more significant than has been appreciated. We argue that this history is shaped by the tension between two sets of epistemic practices, devices and norms. On the one side there were those who emphasised randomisation and sampling to evidence claims about, and justify policies with respect to, the aetiology of disease. On the other side there were those who evoked the technical repertoire of physiological research, especially the notion of the ‘model organism’, to argue for a different integration of the individual in modern society.