1. Divergent evidence suggests that the intracellular signalling pathways for beta-adrenoceptor-mediated vascular relaxation involves either cAMP/protein kinase (PK) A or endothelial nitric oxide (NO) release and subsequent activation of cGMP/PKG. The present study identifies the relative roles of NO and cAMP, as well as dependence on the endothelium for beta-adrenoceptor-mediated relaxation of rat isolated aortas. 2. Cumulative concentration-response curves to isoprenaline (0.01-3 micromol/L) in phenylephrine (0.1 micromol/L)-preconstricted endothelium-intact and -denuded aortas were constructed. Isoprenaline-mediated relaxation was partially reduced by endothelium removal and the presence of the NO synthase inhibitor N(G)-monomethyl-L-arginine (0.1 mmol/L), but not by the cAMP antagonist (Rp)-cyclic adenosine-3',5'-monophosphorothioate (Rp-cAMPS; 0.5 mmol/L). 3. In contrast, in endothelium-denuded aortas, the isoprenaline-mediated relaxation was inhibited by Rp-cAMPS and this inhibition was lost in the presence of the NO donor sodium nitroprusside (1 nmol/L). This effect was not due to phosphodiesterase (PDE) activity because the non-selective PDE inhibitor 3-isobutyl-1-methylxanthine (1 micromol/L) failed to affect the isoprenaline vasorelaxant response. 4. The K(+) channel blocker tetraethylammonium (TEA; 1 mmol/L) attenuated isoprenaline-induced relaxation in endothelium-denuded aorta, but its effect was non-additive with Rp-cAMPS, suggesting that the K(+) channel component may involve cAMP. In endothelium-intact aortas, TEA but not Rp-cAMPS reduced isoprenaline relaxation, suggesting an additional non-cAMP component. 5. These findings suggest that beta-adrenoceptors induce vascular smooth muscle relaxation by acting through the NO-cGMP pathway and, when that is disrupted by endothelium removal or the presence of an NO synthase inhibitor, the cAMP pathway in smooth muscles is used. The lack of cAMP participation in endothelium-intact vessels may be because NO suppresses or overrides the cAMP effect.