1. The involvement of cGMP in vasodilatation produced by a range of nitrodilators was investigated using two different protein kinase G inhibitors, R(p) 8-bromoguanosine-3'5'-cyclic monophosphothioate (RBrcGMPS) and KT 5823. 2. The nitric oxide donors sodium nitroprusside (SNP), glyceryltrinitrate (GTN) and s-nitroso-acetylpenicillamine (SNAP), the endothelium-dependent vasodilator acetylcholine (ACh) as well as the cGMP analogues 8-(4-chlorophenylthio)-cGMP(CPTcGMP) and beta-phenyl-1-N2-etheno-8-bromo-cGMP (PETcGMP) all relaxed rat aortic rings preconstricted with phenylephrine (0.1 micromol/L). 3. The protein kinase G inhibitor KT 5823 (10 micromol/L) produced a very small inhibition of the vasodilatation produced by GTN, but had no effect against vasodilatation produced by SNP, CPTcGMP or PETcGMP, which suggests that KT 5823 is not a useful tool in this system. 4. In contrast, RBrcGMPS (0.5 mmol/L) produced a rightward shift of the concentration-response curves to SNP, CPTcGMP and PETcGMP. RBrcGMPS (0.5 mmol/L) also completely abolished vasodilatation to ACh and GTN but, surprisingly, had no effect on vasodilatation produced by SNAP. 5. The guanylate cyclase inhibitor 1H-[1,2,4] oxadiazolo[4,3-a]quinoxalin-1-one (ODQ; 1 and 10 micromol/L) completely inhibited the relaxation produced by GTN, whereas SNAP still had an appreciable relaxant effect after ODQ (1 micromol/L). 6. The differential effect of RBrcGMPS and ODQ on the nitrodilators suggests that there are differences in the mechanism of dilatation between the nitrodilators.