The intrathecal (i.t.) injection of endothelins to conscious rats was found to cause respiratory arrest. To gain some insights into this central phenomenon, peripheral vascular permeability and lung oedema were measured after i.t. and i.v. injections of these peptides. When injected at T-8 spinal cord level, endothelin-1 (65 and 650 pmol) and endothelin-3 (650 pmol) enhanced vascular permeability in the lungs by 22-fold and 7-fold, respectively, and caused sudden death at the highest dose. Less prominent increases (between 1.4- and 2.2-fold) of vascular permeability were observed in other tissues (trachea, kidney, ears, skin of hind paws and back skin) with endothelin-1. Endothelin-1 (650 pmol) caused a similar increase (27-fold) in lung vascular permeability when injected at T-2, although the response was significantly less (P < 0.05) if injected at the L-4 (15-fold) spinal cord level. Only endothelin-1 produced lung oedema when injected at the T-2 or T-8 level. In contrast, intravenous injection of endothelins-1 and -3 (650 pmol) did not produce lung oedema and the lung vascular permeability was increased by only 1.4-1.6-fold and all rats survived. The prior i.t. injection of 6.5 nmol BQ-123 (cyclo[D-Trp, D-Asp, L-Pro, D-Val, L-Leu]), a selective endothelin ET(A) receptor antagonist, prevented the increases of lung vascular permeability and oedema and the mortality induced by i.t. endothelin-1 (650 pmol). Whereas i.v. treatment with phentolamine (2 mg/kg) or pentolinium (25 mg/kg + 50 mg/kg per h x 15 min) abolished the lung vascular permeability changes evoked by endothelin-1 (650) pmol), atropine (1 mg/kg), NG-nitro-L-arginine (50 mg/kg) or indomethacin (5 mg/kg) had no effect. Moreover, the effects of endothelin-1 were attenuated in capsaicin pretreated rats (125 mg/kg, 10 days earlier) and almost abolished in rats subjected to sympathectomy with 6-hydroxydopamine (100 mg/kg, 24-48 h earlier). All these treatments except atropine and NG-nitro-L-arginine prevented the endothelin-1-induced lung oedema and reduced the lethality by around 50%. These results suggest that the increases of pulmonary vascular permeability and oedema induced by i.t. endothelin-1 are due to an intense pulmonary vasoconstriction mediated by alpha-adrenoceptors following the release of catecholamines in response to the activation of endothelin ET(A) receptor in the spinal cord. This central phenomenon seems to be reflexogenic, including the involvement of primary afferent C-fibers and spinal cord ascending fibers to the brain. Thus, endothelin-1 could play a role in neurogenic pulmonary oedema through a central mechanism.