A highly sensitive radioimmunoassay for cholecystokinin (CCK) without any cross-reactivity with gastrin is described. The antibody was raised in a rabbit by immunisation with 30% CCK and bound to all COOH-terminal CCK-peptides containing at least 14 amino acid residues. The affinity constant of the antibody was 59.4 X 10(10) l/mol. CCK 33 conjugated to [125I]hydroxyphenylpropionic acid-succinimide ester was used as label. The binding between label and antibody was inhibited by 50% (ID50) at a concentration of 2.8 pmol/l cholecystokinin 33. The detection limit of the assay was between 0.5 and 1.0 pmol/l plasma. Concentrations of CCK in aqueous acid extracts of human upper small intestine were 36.5 +/- 9.8 pmol/g and of human cerebral cortex 28.2 +/- 2.5 pmol/g tissue. Plasma samples were extracted in 96% ethanol prior to assay. No advantage was obtained by adding aprotinin to the tubes. When frozen at -20 degrees C plasma CCK was stable for at least 6 months. Basal plasma CCK concentrations in 30 normal subjects were very low, 0.9 +/- 0.1 pmol/l, range 0.5 to 3.1 pmol/l. Intraduodenal administration of fat induced significant increases in plasma CCK from 1.1 +/- 0.1 to 8.2 +/- 1.3 pmol/l (p = 0.01). Infusion of exogenous CCK, resulting in plasma CCK levels slightly lower than those measured during administration of fat, induced pancreatic enzyme secretion and gallbladder contraction. The reliability of this radioimmunoassay for measurements of CCK in human plasma was extensively evaluated.