To be able to perform a comprehensive and rigorous benefit-risk analysis of individual food components, and of foods, a number of fundamental questions need to be addressed first. These include whether it is feasible to detect all relevant biological effects of foods and individual food components, how such effects can confidently be categorised into benefits and risks in relation to health and, for that matter, how health can be quantified. This article examines the last of these issues, focusing upon concepts for the development of new biomarkers of health. Clearly, there is scope for refinement of classical biomarkers so that they may be used to detect even earlier signs of disease, but this approach defines health solely as the absence of detectable disease or disease risk. We suggest that the health of a biological system may better be reflected by its ability to withstand and manage relevant physiological challenges so that homeostasis is maintained. We discuss the potential for expanding the range of current challenge tests for use in conjunction with functional genomic technologies to develop new types of biomarkers of health.