1 In segments of human right atrial appendages and pulmonary arteries preincubated with [3H]-noradrenaline and superfused with physiological salt solution containing desipramine and corticosterone, the involvement of imidazoline receptors in the modulation of [3H]-noradrenaline release was investigated. 2 In human atrial appendages, the guanidines aganodine and DTG (1,3-di(2-tolyl)guanidine) which activate presynaptic imidazoline receptors, inhibited electrically-evoked [3H]-noradrenaline release. The inhibition was not affected by blockade of alpha 2-adrenoceptors with 1 microM rauwolscine, but antagonized by extremely high concentrations of this drug (10 and/or 30 microM; apparent pA2 against aganodine and DTG: 5.55 and 5.21, respectively). 3 In the presence of 1 microM rauwolscine, [3H]-noradrenaline release in human atrial appendages was also inhibited by the imidazolines idazoxan and cirazoline, but not by agmatine and noradrenaline. The inhibitory effects of 100 microM idazoxan and 30 microM cirazoline were abolished by 30 microM rauwolscine. 4 In the atrial appendages, the rank order of potency of all guidelines and imidazolines for their inhibitory effect on electrically-evoked [3H]-noradrenaline release in the presence of 1 microM rauwolscine was: aganodine > or = BDF 6143 [4-chloro-2-(2-imidazolin-2-yl-amino)-isoindoline] > DTG > or = clonidine > cirazoline > idazoxan (BDF 6143 and clonidine were previously studied under identical conditions). This potency order corresponded to that previously determined at the presynaptic imidazoline receptors in the rabbit aorta. 5 When, in the experiments in the human pulmonary artery, rauwolscine was absent from the superfusion fluid, the concentration-response curve for BDF 6143 (a mixed alpha 2-adrenoceptor antagonist/imidazoline receptor agonist) for its facilitatory effect on electrically-evoked [3H]-noradrenaline release was bell-shaped. In the presence of 1 microM rauwolscine, BDF 6143 and cirazoline concentration-dependently inhibited the evoked [3H]-noradrenaline release. 6 In human atrial appendages, non-adrenoceptor [3H]-idazoxan binding sites were identified and characterized. The binding of [3H]-idazoxan was specific, reversible, saturable and of high affinity (KD: 25.4 nM). The specific binding of [3H]-idazoxan (defined by cirazoline 0.1 mM) to membranes of human atrial appendages was concentration-dependently inhibited by several imidazolines and guanidines, but not by rauwolscine and agmatine. In most cases, the competition curves were best fitted to a two-site model. 7 The rank order of affinity for the high affinity site (in a few cases for the only detectable site; cirazoline = idazoxan > BDF 6143>DTG> or = clonidine) is compatible with the pharmacological properties of I2-imidazoline binding sites, but is clearly different from the rank order of potency for inhibiting evoked noradrenaline release from sympathetic nerves in the same tissue. 8 It is concluded that noradrenaline release in the human atrium and, less well established, in the pulmonary artery is inhibited via presynaptic imidazoline receptors. These presynaptic imidazoline receptors appear to be related to those previously characterized in rabbit aorta and pulmonary artery, but differ clearly from I1 and I2 imidazoline binding sites.