This study examined the relative importance of five risk factors and health behaviours (namely dietary habits, leisure time exercise, smoking, alcohol consumption and body mass index) on self-ratings of health among the Swedish adult population. The data come from the 1991 Swedish Level of Living Survey, a face-to-face survey interview based on a sample representative of the Swedish population aged between 18 and 75 years (n = 5306). The analyses were carried out using logistic regression analysis. With the exception of the consumption of dietary fat, all the risk factors and health behaviours studied were associated with self-rated health. When they were adjusted for health problems and functional limitations most of the associations weakened or disappeared altogether, but smoking and use of vegetables in the diet were still associated with self-rated health. Self-ratings of young adults (18-34 years) were found to be related to body mass index even when health problems were adjusted for, with both obesity and underweight contributing to less than good self-rated health. The results indicate that risk factors and health behaviours do not, in general, directly contribute to self-ratings of health. Instead, their effect is mediated by more specific health problems and their functional consequences. However, smoking and not consuming vegetables, as well as obesity and underweight among young respondents, were found to have an independent association with self-rated health. This may reflect the effects of health problems not captured by our indicators of ill health, but may also indicate that risk factors and risky behaviours are considered to have an effect on one's perceived health even in the absence of health consequences.