The consequences of fetal growth retardation remain unclear, in part because they appear to vary between industrialized and developing countries. Data on body composition offer a new opportunity to investigate this issue, and may be of particular value in addressing the controversial role of nutrition in infancy, which has been proposed by some to boost survival, and by others to increase long-term risk of chronic diseases. The uncertainty regarding the effects of post-natal nutrition is presenting challenges to nutritional policy as many countries undergo the nutrition transition, whereby the nutritional status of individuals may shift within the life-course. A theoretical model, building on the thrifty phenotype hypothesis, is presented to clarify how body composition data can address this dilemma. Measurements of body composition can now be obtained in infants and children using several different technologies, indicating that large-scale studies can now be conducted to investigate objectively the association between early growth patterns and later health.