1 Although adenosine analogues such as 5'-N-ethylcarboxamidoadenosine (NECA) relax the rat thoracic aorta in a partially endothelium-dependent manner via adenosine A 2A receptors, others such as N 6 -R-phenylisopropyladenosine (R-PIA) act via an endothelium-independent, antagonistinsensitive mechanism. The role of cyclic nucleotides in these relaxations was investigated in isolated aortic rings using inhibitors of adenylate and guanylate cyclases as well as subtype-selective phosphodiesterase inhibitors. 2 The adenylate cyclase inhibitor 9-(tetrahydro-2-furanyl)-9H-purin-6-amine (SQ 22536; 100 mM) signi®cantly inhibited responses to NECA, but not responses to R-PIA. The type IV (cyclic AMPselective) phosphodiesterase inhibitor 4-[(3-butoxy-4-methoxyphenyl)methyl]-2-imidazolidinone (RO 20-1724; 30 mM) signi®cantly enhanced responses to NECA and to a lesser extent those to R-PIA. 3 The guanylate cyclase inhibitor 1H-[1,2,4]oxadiazolo [4,3a]quinoxalin-1-one (ODQ; 100 mM) signi®cantly inhibited responses to NECA and acetylcholine but not responses to R-PIA. The selective phosphodiesterase V (cyclic GMP-selective) inhibitors, zaprinast (10 mM) and 4-{[3',4'-(methylenedioxy)benzyl]amino}-6-methoxyquinazoline (MMQ; 1 mM), had no signi®cant e ect on responses to either NECA or R-PIA, but enhanced responses to acetylcholine. 4 These results are consistent with the e ects of NECA being via activation of endothelial receptors to release NO which stimulates guanylate cyclase, as well as smooth muscle receptors coupled to stimulation of adenylate cyclase. The lack of e ect of zaprinast and MMQ on responses to NECA are likely to be due to simultaneous activation of both adenylate and guanylate cyclases in the smooth muscle, as cyclic AMP reduces the sensitivity of phosphodiesterase V to inhibitors. These results also suggest that the e ects of R-PIA are via neither of these mechanisms.