The first half of this manuscript is devoted to a review of the methods used and the results obtained in the published measurements of the normal responses to tests of the three main types of hypothalamic-pituitary-adrenocortical (HPA) activity in man. These are, I, basal, unstressed activity leading to appropriate levels of total daily production of cortisol in the characteristic circadian pattern; II, responses to feedback stimulation of HPA activity by metyrapone administration; and III, responses to tests of the effects of stress on the HPA system including the effects of hypoglycemia, induced fever, vasopressin administration, and ACTH injections and infusions. The advantages and shortcomings of each type of procedure are discussed. The second half of this paper describes the authors' attempts to establish the limits of normality of standard and modified methods of evaluating the HPA system. The defined limits of normality have been used to assess the HPA function in 158 patients with known or suspected disorders of the HPA system. In normal controls, halfhourly plasma cortisol determinations established the normality of circadian and postprandial fluctuations and of mean plasma cortisol concentration, 6.2 +/- 0.3 (SEM) micrograms/dl, which were closely approximated by determinations every 6 h. Metyrapone, given in a dose of 500 mg every 2 h for 24 h increased urinary 17-OHCS excretion to 10.5-32.6 mg/day or to 1.7-7.8 times basal excretion rate. Increasing rates of insulin infusion disclosed significant relationships between resulting plasma glucose and cortisol concentrations. The slopes of the delta cortisol/delta glucose responses were similar after insulin infusions (0.46 +/- 0.05) and after insulin injections, 0.15 U/kg (0.43 +/- 0.09), and were always greater than 0.20 micrograms/mg. This index provides a useful objective measure of the normality of responses to hypoglycemic stress, 0.20-0.87 micrograms/mg. Adrenocortical responses to iv infusions of ACTH (cosyntropin 0.25 mg) may be equivocal at 2 h but are clear cut at 4, 6 and 8 h. Of 158 patients in whom hypopituitarism was known or suspected because of the presence of a pituitary tumor, acromegaly, hyperprolactinemia, or clinical features, HPA function was found to be entirely normal in 88 patients and partially or severely abnormal in the remaining 70 patients.(ABSTRACT TRUNCATED AT 400 WORDS)