), vasodilatory effects of CGRP are mediated via the endothelium and are due to the production of nitric oxide. In the majority of vessels studied to date, however, CGRP causes vasodilatation through an endothelium-independent process, not involving nitric oxide. These vessels include the rat mesenteric, gastric, splenic and renal arteries Li & Duckles, 1992;Amerini, Mantelli & Ledda, 1993;Holzer, Lippe, Jocic, Wachter, Erb & Heinemann, 1993;Gao, Nishimura, Suzuki & Yoshida, 1994), rabbit jejunal (La & Rand, 1993), rat and rabbit hepatic artery Brizzolara & Burnstock, 1991), skin and skeletal muscle microcirculations (Persson, Hedqvist & Gustafsson, 1991;Ralevic, Khalil, Dusting & Helme, 1992;Brain, Hughes, Cambridge & O'Driscoll, 1993) dog basilar (Oyama et al. 1993), cat cerebral (Saito, Masaki, Uchiyama, Lee & Goto, 1989), dog lingual (Kobayashi, Todoki, Ozono & Okabe, 1995) and human uterine arteries (Bodelsson & Stjernquist, 1992 1. Arteriolar diameter and membrane voltage have been measured to investigate the actions of calcitonin gene-related peptide (CGRP) in rat irideal arterioles. 2. Activation of sensory nerves inhibited sympathetic vasoconstriction, reduced the accompanying 40-50 mV depolarization by 90 % and caused a 4 mV hyperpolarization. 3. The inhibition of vasoconstriction was prevented by either preincubation in ¬_NAME (10 ìÒ), to inhibit nitric oxide production, by preincubation in the cell-permeant adenylate cyclase inhibitor dideoxyadenosine (1 mÒ) or by preincubation in the ATP-sensitive potassium channel blocker glibenclamide (10 ìÒ). The subsequent addition of a nitric oxide donor to the glibenclamide solution inhibited nerve-mediated vasoconstriction, suggesting that the potassium channel involvement preceded the production of nitric oxide. The small hyperpolarization was not affected by ¬_NAME. 4. Nerve-mediated vasodilatation persisted in the presence of ¬_NAME (10 ìÒ) but was abolished with the CGRPÔ receptor antagonist CGRP8-37. 5. In arterioles preconstricted with the áµ-adrenoceptor agonist UK-14304 (100 nÒ), exogenous CGRP caused a hyperpolarization and a dose-dependent vasodilatation, neither of which was affected by ¬_NAME (10 ìÒ). 6. In arterioles preconstricted with 30 mÒ KCl, CGRP (10 nÒ) caused vasodilatation but not hyperpolarization, suggesting that the hyperpolarization was not causal to the vasodilatation. 7. Forskolin (30 nÒ), in the presence of ¬_NAME to prevent effects due to nitric oxide, caused vasodilatation. 8. These results suggest that CGRP inhibits sympathetic nerve-mediated vasoconstriction through sequential increases in cyclic AMP and nitric oxide, while vasodilatation results from increases in cyclic AMP alone. The production of nitric oxide, but not its mechanism of action, appears to be dependent on the activation of ATP-sensitive potassium channels. The possible sites of action of these two pathways are discussed.