This paper reviews child and adolescent adiposity measures and associated long-term health risks. The first section argues that anthropometric measures are practical for large scale epidemiological studies, particularly the body mass index. Limitations of this and other measures are presented. The second section summarises the evidence on the relationship between child and adolescent and adult adiposity. This is based on a search for relevant literature in the following computerised databases: Medline (1985-96), BIDS (EMBASE and Science Citation Index 1985-96). The literature search revealed that the child to adult adiposity relationship is now well-documented, although methodological differences hinder comparisons. Nonetheless, consistently elevated risks of adult obesity are evident for fatter children, although the prediction of adult obesity from child and adolescent adiposity measures is only moderate. Fewer studies could be identified in relation to long-term health risks of child and adolescent adiposity. It is therefore difficult to specify categories of risk associated with childhood adiposity without more information from long-term studies. Further evidence is also required to confirm the suggestion from some studies that adult disease risks are associated with a change in adiposity from normal weight in childhood to obesity in adulthood. However, on the basis of the evidence available, it is argued that population-based approaches to the prevention of obesity are likely to be more effective than approaches targeted as fat children. Population-based approaches are desirable, first because of the poor prediction of adult obesity from child and adolescent measures, and second, because of risks of adult mortality and morbidity may be elevated for individuals who become overweight after adolescence.