Summary1. The characteristic histological feature of the pars nervosa is a dense network of nonmyelinated nerve fibres, which end within the pars nervosa and not in relationship to glandular cells of the pituitary. These fibres originate in the supraoptic and paraventricular nuclei of the hypothalamus and pass through the median eminence into the pituitary stalk.2. The antidiuretic, pressor and oxytocic activities of extracts of the posterior lobe of the pituitary are derived from the pars nervosa and not from the pars intermedia. Despite similarities in anatomical structure between the tissues of the tuber cinereum, pituitary stalk and pars nervosa, extracts of stalk and tuber cinereum have very little pharmacological activity, so that the only established source of pituitary antidiuretic substance is the pars nervosa itself.3. The antidiuretic action of extracts of the posterior lobe is described, and it is emphasized that this action is only demonstrable when there is rapid secretion of dilute urine, as in water diuresis in the intact animal, in diabetes insipidus and in perfused kidneys. The antidiuretic action is produced by very small amounts of the extract.4. In order to prove that the antidiuretic substance is a true hormone, the evidence for its release into the circulation during life is examined. The tests available for the detection of antidiuretic activity are not adequate to allow the demonstration of antidiuretic substance in the body fluids; but it has been conclusively established that the inhibition of water diuresis which may occur in the unanaesthetized dog from emotional stress or from the intravenous injection of acetylcholine, morphine or nicotine is mediated by the release of antidiuretic hormone from the pars nervosa.5. It has long been believed that clinical diabetes insipidus is due to deficiency of antidiuretic hormone; in sT V the results on the fluid exchange of experimental lesions in the pituitary region are examined. Permanent diabetes insipidus regularly results from section of the supraoptico‐hypophyseal tracts in the median eminence, and as the result of the section there is atrophy and loss of function of the pars nervosa. However, operative removal of the posterior lobe causes only a temporary increase in the fluid exchange, but permanent diabetes insipidus has resulted in some instances where the posterior lobe, stalk and part of the tuber cinereum were removed. It is usually accepted that the stalk and tuber cinereum as well as the pars nervosa are significant sources of the antidiuretic hormone, and that diabetes insipidus only occurs when these three parts of the neurohypophysis are removed or atrophied, but there is only limited evidence in support of this explanation. The relation of the anterior lobe to the production of diabetes insipidus is also discussed.6. Thus, there are three lines of evidence which justify the acceptance of the pars nervosa as controlling the rate of urine secretion by means of its antidiuretic hormone: the antidiuretic activity of extracts, the release of antidiuretic hormone during emotional stress, and the association of diabetes insipidus with loss of the function of the pars nervosa.7. The evidence is inadequate to‐establish other suggested functions of the mammalian pars nerv