Glucocorticoid replacement is essential in patients with primary and secondary adrenal insufficiency, but many patients remain on higher than recommended dose regimens. There is no uniformly accepted method to monitor the dose in individual patients. We have compared cortisol concentrations in plasma, saliva and urine achieved following "physiological" and "stress" doses of hydrocortisone as potential methods for monitoring glucocorticoid replacement. Cortisol profiles were measured in plasma, saliva and urine following "physiological" (20 mg oral) or "stress" (50 mg intravenous) doses of hydrocortisone in dexamethasone-suppressed healthy subjects (8 in each group), compared to endogenous cortisol levels (12 subjects). Total plasma cortisol was measured half-hourly, and salivary cortisol and urinary cortisol:creatinine ratio were measured hourly from time 0 (between 0830 and 0900) to 5 h. Endogenous plasma corticosteroid-binding globulin (CBG) levels were measured at time 0 and 5 h, and hourly from time 0 to 5 h following administration of oral or intravenous hydrocortisone. Plasma free cortisol was calculated using Coolens' equation. Plasma, salivary and urine cortisol at 2 h after oral hydrocortisone gave a good indication of peak cortisol concentrations, which were uniformly supraphysiological. Intravenous hydrocortisone administration achieved very high 30 minute cortisol concentrations. Total plasma cortisol correlated significantly with both saliva and urine cortisol after oral and intravenous hydrocortisone (P <0.0001, correlation coefficient between 0.61 and 0.94). There was no difference in CBG levels across the sampling period. An oral dose of hydrocortisone 20 mg is supraphysiological for routine maintenance, while stress doses above 50 mg 6-hourly would rarely be necessary in managing acute illness. Salivary cortisol and urinary cortisol:creatinine ratio may provide useful alternatives to plasma cortisol measurements to monitor replacement doses in hypoadrenal patients.