The present report discusses the main constraints influencing the choice in the method in the epidemiological approach of host factors and family factors. It briefly describes 3 of the usual methods in this field, i.e. the migrant studies, the twin studies and the family studies. The practical problems and the conceptual limitations met by each of these 3 methods are reviewed. A brief review of the literature provides an opportunity for examining the main misunderstandings between geneticists and epidemiologists in their study of human groups when analysing the role of such factors in the etiology of cancer.