The mammalian conceptus is surrounded by, and entirely supported by, maternal extracellular fluid, an environment whose stability is well regulated by the mother's own homeostatic mechanisms. However, it appears to be necessary for a fetus to maintain some independence from its mother. The placenta provides such a barrier; the fetus can enjoy a separate 'milieu interieur', which may be an advantage from immunological, endocrinological or nutritional aspects.Animal experiments suggest that the placental barrier is epithelial in nature. In common with other epithelia, it is relatively permeable to substances of high lipid solubility, such as gases. There is good evidence for the presence of special transfer mechanisms for those lipid-insoluble substances of biological importance such as sugars, aminoacids and at least some electrolytes. The development of such transfermechanisms would have been an evolutionary necessity in the presence of an otherwise relatively impermeable barrier, as transfer of adequate quantities of anabolic substrates needs to be guaranteed. Little is yet known of the selectivity of these mechanisms or of what controls them - or if they are controlled at all; and there is no evidence concerning whether their failure contributes to materno-fetal disease. At present, the placental epithelium may be considered to have a transport function in search of a pathology.Most experimental data have been obtained in animals, but observations in man suggest that, while there are differences in detail, there is considerable similarity in the overall pattern of placental epithelial function. Experiments made by man occur whenever medication is prescribed. The different patterns of permeability to drugs of differing properties provide support for the relative impermeability to lipid insoluble molecules, while the neonatal hyponatremia following intravenous glucose infusions to women in labour reflects the high placental permeability to water.Experiments of nature often depend on the constitutional differences between mother and the genetically different fetus. The markedly different protein phenotypes in mother and fetus implies a very low passive permeability to protein. Similarly the relative impermeability to thyroxine is evident from the delayed osseous maturation of the hypothyroid fetus in spite of normal maternal thyroxine levels. In some cases placental transfer functions contribute to fetal disease; for example, the placenta fails to protect the fetus from the hostile environment of maternal diabetes, phenylketonuria or rhesus isoimmunisation.Most such observations pose more questions than they answer. Yet these answers must be sought if we are fully to understand, and perhaps modify, the contribution of placental transfer mechanisms to fetal health and disease.