In most clinical situations, the ability of the adrenal to produce cortisol is studied with measurements of cortisol concentration. The validity of this assumes the existence of a predictable relationship between cortisol production and plasma cortisol concentration. This assumption was tested by determining if production rates, as simulated by constant cortisol infusions, were predictive of the resultant plasma and urine cortisol concentrations in dexamethasone-suppressed subjects. Additional attempts to predict simulated cortisol production rates were made with infusions of [9,12,12-2H3]cortisol. Thirty-six 8-h cortisol infusions were performed in 24 subjects. Although there was a tendency to achieve a higher plasma cortisol concentration response at a higher cortisol infusion rate (P < 0.04), the known cortisol infusion rate was a poor predictor of plasma cortisol responses (r2 = 0.12). Addition of corticosteroid-binding globulin concentration (CBG) as a covariate greatly improved the predictability of plasma cortisol responses (r2 = 0.52). Urine free cortisol, normalized to inulin excretion, had a predictable relationship to cortisol infusion rate (r2 = 0.82). Isotope-enrichment data from a group of 6 subjects receiving multiple cortisol infusions during administration of [9,12,12-2H3]cortisol gave the most accurate predictions of the known cortisol infusion rates (r2 = 0.93). It is concluded that: 1) cortisol infusion (production) rate and plasma cortisol concentration are poorly correlated; 2) CBG is an important modulator of concentration responses to a given production rate; and 3) short (8-h) infusions of [9,12,12-2H3]cortisol may be used to determine cortisol production rates in humans. The mechanisms by which CBG modulates cortisol concentration responses are elaborated in a companion report.