ACTH tests were performed with and without dexamethasone (dex) pretreatment to clarify the nature of the relationship between the absolute and incremental response (delta) to ACTH in normal men and women, hirsute women, adrenarchal children, and women heterozygous for congenital adrenal hyperplasia (CAH). The purposes were to test the effect of dex preparation on adrenal responsiveness to ACTH and the efficacy of the dex-pretreated ACTH test in detecting heterozygosity for CAH. Cosyntropin was given as a 10 micrograms/m2 iv bolus dose at 0800-1000 h; dex (1 mg/m2) was given at 2200-2400 h the night before. Dex did not alter the absolute plasma steroid levels achieved in response to ACTH. However, since post-dex baseline concentrations of adrenal steroids were lower, the delta to ACTH was significantly greater for the major adrenal secretory products, 17 alpha-hydroxypregnenolone (3 beta, 17 alpha-dihydroxypregn-5-ene-20-one), dehydroepiandrosterone, and cortisol (F). For example, for all paired tests, the mean plasma F values achieved 30 min post-ACTH were 26.0 +/- 4.4 (+/- SD) micrograms/dl without dex and 23.8 +/- 5.5 micrograms/dl after dex. In contrast, the mean delta of plasma F 30 min post-ACTH was less without (13.3 +/- 4.8 micrograms/dl) than after (19.4 +/- 3.3 micrograms/dl) dex (P less than 0.001). Apparent 21-hydroxylase efficiency, computed from dex-prepared tests, was found in follicular phase women to have a markedly skewed distribution without clear demarcation between 15% of the population and CAH heterozygotes. Luteal phase responses differed from follicular phase responses in dex-pretreated women in the magnitude of the 17-hydroxyprogesterone response. In the luteal phase, although the plasma 17-hydroxyprogesterone level at 30 min was higher, a response to ACTH was not consistently found, averaging only 22 +/- 26 ng/dl, in contrast to the consistent 52 +/- 15 ng/dl response in the follicular phase. These findings have practical implications for interpreting rapid ACTH test results. The absolute plasma F level achieved post-ACTH is more important as an index of adrenocortical reserve than the increment. Dex pretreatment appears to offer no practical advantage in ACTH testing for mild defects in 21-hydroxylation; we postulate that this is because of considerable normal variability in the efficiency of 21-hydroxylation.