At the beginning of the 20th century, and more particularly in the decades following the Second World War, the methodological formalisation and the increasingly frequent use of case-control and prospective cohort study designs played a major role in the development of etiological research in epidemiology. The history of the prospective cohort study is best known for the study of cancer epidemiology ; with this design being developed and improved in the context of an animated controversy over the causal status of cigarette smoking as a risk factor in lung cancer. In this paper, I focus on the history of this kind of study design in American cardiovascular epidemiology, and, more particularly, the Framingham heart study. This study rapidly came to be considered as the prototype and model for this method of study and an icon of its potential success in etiological research in cardiovascular epidemiology. A comparison with cancer epidemiology reveals the differences between this type of study in the context of cardiovascular disease in terms of both the approach and its origins. I shall first highlight the main differences and specificities of the Framingham study in terms of the nature of its cohort and the followup. I will then show how this cohort study design has some roots both in longitudinal studies led on general populations by the epidemiologists of the United States Public Health Service at the beginning of the 20th century, and in « follow-up studies » conducted by clinicians.