SUMMARYWe have developed a sensitive two‐site immunoradiometric assay (IRMA) for intact ACTH and its precursors, pro‐opiomelanocortin and 22 kDa peptide in unextracted human plasma. The assay uses two monoclonal antibodies. Antibody 1A12, specific for ACTH 10–18, is radiolabelled and antibody 2A3 specific for the C‐terminal region (ACTH 24–39), is coupled to Sephacryl S300 for the solid‐phase. Samples are incubated for 18 h with labelled antibody followed by 2 h with solid‐phase antibody. Separation employs the sucrose layering technique. Using human pituitary ACTH 1–39 (code 74/555) in diluent containing 10% horse serum to standardize the assay, the sensitivity (upper 99% confidence limit of zero standard) is 3.5 ± 0.8 ng/l (n= 7). The mean coefficient of variation is 5.9% within‐assay and 6.7% between‐assay and is less than 10% between 22 and>5000 ng/l. Mean recovery of ACTH 1–39 added to dexamethasone‐suppressed human plasma is 109% and endogenous ACTH behaves indistinguishably from standard ACTH on dilution. In normal subjects, mean plasma ACTH levels are 30 ng/l at 0730 h, and 15 ng/1 at 1630 h at rest. ACTH concentrations are between 60 and 330 ng/l, 8–10.5 h after metyrapone (2 g orally at 2300 h), between 140 and 320 ng/1, 30–60 min after insulin‐induced hypoglycaemia, and<4 ng/1, 8 h after dexamethasone (1.5 mg orally at 2300 h). In a range of pathological conditions ACTH concentrations accurately reflect the disorders of the pituitary‐adrenal axis. Endogenous ACTH immunoactivity is stablein vitroat 22°C for at least 1 h in whole blood and at least 4 h in plasma. It is concluded that this two‐site IRMA for ACTH in unextracted plasma offers a r