The glucagon stimulation test (GST) is often used to assess pituitary ACTH reserve, particularly when other tests are contra-indicated. In a preliminary investigation, in patients with pituitary disease, we failed to demonstrate the ACTH dependence of the cortisol response. We have therefore compared the ACTH, cortisol and glucose responses to glucagon (1 mg s.c.), insulin (0.2 U/kg i.v., ITT) and placebo in six healthy male volunteers, sampling every 10 min for 6 h. During the GST, mean +/- SD serum cortisol rose from 256 +/- 80 nmol/l to a peak of 481 +/- 164 nmol/l (range 289-717 nmol/l, P less than 0.01) in comparison with 280 +/- 81 nmol/l to 602 +/- 110 nmol/l (range 493-742 nmol/l) during the ITT (P less than 0.002). The mean peak cortisol levels achieved in the two tests did not differ significantly. In the GST, plasma ACTH rose from a mean basal value of 10.9 +/- 16.6 ng/l to a mean peak level of 123 +/- 76 ng/l (P less than 0.02) (ACTH ng/l x 0.225 = pmol/l). The corresponding values in the ITT were 7.1 +/- 16.2 ng/l and 263 +/- 91 ng/l (P less than 0.001). The mean peak ACTH level was significantly greater during the ITT (P less than 0.05). Thus the cortisol response was ACTH dependent in both the GST and the ITT in normal subjects. Furthermore, the ACTH response was of sufficient duration to be detected by the usual procedure of sampling every 30 min.(ABSTRACT TRUNCATED AT 250 WORDS)