Few data are available on the reliability of measurements of adrenocortical and corticotroph hormones for use in clinical pharmacology. Two placebo controlled cross-over trials in 20 normal healthy male subjects offered the opportunity to perform three repeat samplings of adrenocortical and corticotroph hormones at 1 to 5 week intervals during the placebo periods. Measurements of baseline levels of plasma, salivary and urinary cortisol, plasma adrenocorticotroph hormone (ACTH), lipotrophic hormone (LPH), beta-endorphin, post tetracosactrin levels of plasma and salivary cortisol, post corticotrophin releasing hormone (CRH)-lysine vasopressine (LVP) levels of plasma cortisol, ACTH and LPH; and post metyrapone levels of plasma cortisol and 11-deoxycortisol (compound S), ACTH, LPH, beta-endorphin were performed in the same laboratory. The reliability of the measurements was estimated by computing the intraclass correlation coefficient (R) and by using Altman-Bland graphical method. The Rs of baseline parameters varied from 0.18 (for 08.00 h salivary cortisol) to 0.55 (for 08.00 h plasma cortisol and nocturnal urinary cortisol). In contrast, parameters obtained after direct stimulation or inhibition of the producing targets were much more reliable: Rs were above 0.80 for post tetracosactrin levels of plasma and salivary cortisol, post CRH-LVP levels of plasma ACTH and LPH. The Rs were below 0.50 for post metyrapone levels of plasma 11-deoxycortisol, ACTH, LPH and beta-endorphin. The interval between sampling did not affect R estimates. These data show that peak levels of plasma cortisol and ACTH after direct stimulation are highly reliable whereas baseline and main post-metyrapone levels are not.(ABSTRACT TRUNCATED AT 250 WORDS)