The epidemiological links observed between fetal and infant growth and impaired glucose tolerance in adult life that led to the formulation of the “thrifty phenotype” hypothesis have been confirmed by others in widely differing populations. The proposed nutritional basis of these links has been tested in an animal model in which rat dams were fed an isocaloric low-protein diet and the postwearing normally fed offspring were studied. Permanent changes in key hepatic enzymes of glycolysis and gluconeogenesis and reductions in both insulin and glucagon sensitivity of the liver have been observed. Glucose tolerance deteriorated more at age 15 months compared with controls but animals were not frankly diabetic. The longevity of offspring was affected positively when growth was retarded by postnatal protein restriction, but negatively when protein restriction during pregnancy was followed by cross-fostering of pups to normally fed dams with consequent “catch-up” growth. These effects were greatest in males. The thrifty phenotype hypothesis continues to provide a useful conceptual and experimental basis for the study of the aetiology of non-insulin-dependent diabetes.