There is an extensive literature within anthropology, sociology and psychology about lay concepts of determinants of health and illness. Many of these studies have used single sex or social class samples, often in narrow age bands, and many are qualitative in approach. We asked respondents in a health survey to say how important (on a five-point scale) they thought seven potential influences on health (habits, self-care, the environment, family relationships, one's constitution, money and luck) were. The first three were regarded as very important, the second three as less important and luck as least important. Responses were consistent with current public health and epidemiological knowledge; these respondents endorsed prevailing views about personal responsibility for health and about the role of the physical and social environment in influencing health. In mutually adjusted models, there were no significant gender differences, social class differences and neighbourhood differences in three out of seven influences, and age differences in four out of seven influences. Thus, socio-demographic differences were less marked than might be inferred from studies of specific social groups, indicating a need for caution in health education and health promotion practice against always assuming socio-demographic differences. (Contains 2 tables and 1 figure.)