This study examines whether the assumptions embedded in nutrigenomics, especially the alleged relation between information about personal health risks and healthy behaviour, match how people account for the relation between food, health and genes in everyday life. We draw on discourse analysis to study accounts of overweight in six group interviews with people who are and who are not overweight. The results show potentially contradictory normative orientations towards behavioural explanations of (over)weight. Overt gene accounts are interactionally problematic (in contrast to more indirect accounts such as ‘build’), indicating that participants treat ‘behaviour’ as the normatively appropriate explanation for overweight. At the same time, however, healthy behaviour is an accountable matter, i.e. it is dealt with in interaction as behaviour that is not self-evidently right but requires an explanation. It is discussed how bringing these interactional concerns to the surface is essential for understanding future users’ response to nutrigenomics and emergent technologies more in general.