Variables measuring aspects of social support and/or social network have been correlated with morbidity and mortality in multiple investigations. However, the findings from these investigations are inconsistent. Both conceptual and methodological issues in the literature on social support and health need to be resolved in order to clarify the health promotion implications of this subject. This paper discusses issues in the literature on social support and health, and presents findings relevant to these issues from a survey investigation of illness and health related behaviour in Denmark. Evidence was found of both direct and indirect effects of a non-supportive social network on illness. The effects were not due to the amount of social contact or the quantity of social relationships as measured in this study, but to the extent to which a supportive network reduces feelings of isolation and provides support and help when needed. Findings regarding health-related behaviour point to the fact that personal behavioural patterns (especially habits of coping with tension or stress-producing problems) and their interaction with influences from the social environment have not been modelled together in health research. The manner in which social support is used to avoid or reduce stress-producing problems in contrast to health-damaging habits of stress reduction is a potentially fruitful area of inquiry.