Abstract Negative feedback systems are believed to be operative in higher animals in physiologically important processes ranging from formation of tropic hormones to maintenance of organ size. However, model systems for the study of such controls at the molecular level in intact animals are quite rare, and are limited to two biosynthetic pathways: the cholesterol-bile acid pathway, and the creatine pathway. In this paper the properties of the latter model system are described in detail. Two enzymes are involved in the biosynthesis of creatine. The second enzyme appears to be constitutive, whereas the steady-state level of the first enzyme, arginine-glycine transamidinase, is responsive to the tissue concentration of creatine. This process has been operationally termed end-product repression, by analogy with bacterial systems, until its mechanism can be more completely elucidated. Creatine repression of transamidinase has been observed in the rat, mouse, rabbit, chick, and duck, in tissues as diverse as kidney, pancreas, and liver. More recently, repression has been studied in the liver of the developing chick embryo and the newly hatched chick. Virtually complete repression of embryonic liver transamidinase can be maintained throughout development by a single injection of creatine into the egg. Derepression occurs in the first week following birth, when chicks are fed a normal diet. Numerous experiments have shown that there is a highly specific relationship between the target enzyme and the controlling compound. Creatine precursors proximal to the target enzyme repress 35–50 per cent while the precursor distal to the target enzyme represses completely. In the closed system of the egg, repression of the target enzyme can be readily shown to be proportional to repressor concentration. Moreover, this system permits the demonstration that repression can occur in the absence of intestinal flora, and under conditions of minimal hormonal influences. The normal pattern of change of transamidinase activity during embryonic and neonatal development is consistent with a repression by endogenous creatine prior to birth, followed by a post-hatch derepression, but other explanations are also entertained. Evidence is cited in support of the thesis that the creatine-transamidinase control system has survival value for birds, and perhaps reptiles and amphibians. It is suggested that liver transamidinase of carnivorous birds is normally in a partially repressed state, as a result of the 0·4 per cent creatine content of ingested muscle tissue, whereas transamidinase of herbivorous birds is normally derepressed. Experimentally it has been demonstrated that fasting lowers the activity of both the repressed and derepressed enzymes. At least part of this decrease can be attributed to a repression by endogenous creatine which appears in increased concentration in the blood, liver and kidneys of most higher animals during fasting. During fasting, then, the decrease in transamidinase activity permits diversion of a portion of the dietary essential amino acids, arginine, glycine, and methionine, from the synthesis of creatine, now in excess, to more immediately essential biosyntheses. For example, glycine is essential for synthesis of the uric acid required to remove amino groups arising from the gluconeogenesis of fasting; methionine methyl groups are required for the increased lipid transport of fasting; and all three amino acids are needed for synthesis of essential enzymes, protein hormones, and feathers. In addition to the foregoing, the implications of the occurrence of a repressible system during embryonic development for the problem of the establishment and maintenance of tissue-specific enzyme levels are discussed.