1. The effects of selective inhibitors of adenosine 3':5'-cyclic monophosphate (cyclic AMP) and guanosine 3':5'-cyclic monophosphate (cyclic GMP) phosphodiesterases (PDEs) were investigated on PDEs isolated from the rat aorta and on relaxation of noradrenaline (1 microM) precontracted rat aortic rings, with and without functional endothelium. 2. Four PDE forms were isolated by DEAE-sephacel chromatography from endothelium-denuded rat aorta: a calmodulin-activated PDE (PDE I) which hydrolyzed preferentially cyclic GMP, two cyclic AMP PDEs (PDE III and PDE IV) and one cyclic GMP-specific PDE (PDE V). The latter was selectively and potently inhibited by zaprinast. The two cyclic AMP PDEs were discriminated by specific inhibitors: one was inhibited by cyclic GMP (PDE III) and by new cardiotonic agents (milrinone, CI 930, LY 195115 and SK&F 94120); the other was inhibited by denbufylline and rolipram (PDE IV). None of these drugs significantly inhibited PDE I. 3. The PDE III inhibitors caused endothelium-independent relaxations of rat aortic rings with the following EC50 values (microM concentration producing 50% relaxation): LY 195115: 3.4, milrinone: 5.7, CI 930; 7.8, SK&F 94120: 14.7. Neither NG-monomethyl-L-arginine (L-NMMA, 300 microM), an inhibitor of the L-arginine-NO pathway, nor L-arginine (1 mM) modified the effect of PDE III inhibitors. However, methylene blue (10 microM) an inhibitor of soluble guanylate cyclase abolished relaxation induced by PDE III inhibitors except in the case of compound CI 930. 4. The specific PDE IV and PDE V inhibitors both produced endothelium-dependent relaxations which were inhibited by L-NMMA and by methylene blue (10 microM). In the presence of L-NMMA, relaxation was restored by subsequent addition of L-arginine. 5. The relaxant effects of denbufylline and rolipram were studied in the presence of drugs stimulating either adenylate cyclase (forskolin and isoprenaline) or soluble guanylate cyclase (sodium nitroprusside, SNP), or inhibiting PDE III (milrinone). In endothelium-denuded rings, a relaxing effect of both denbufylline and rolipram was found in the presence of milrinone (EC5o values 1.7 and 12 microM, respectively) or SNP (EC50 values 12.3 and 124 microM, respectively), but not in the presence of forskolin or isoprenaline. However in the presence of functional endothelium, relaxations produced by PDE IV inhibitors were significantly potentiated by forskolin, isoprenaline, milrinone and SNP (respective EC50 values for denbufylline: 2, 2, 0.4 and 0.7 microM and for rolipram: 7, 13, 7 and 1.2 microM). 6. These results indicate that the relaxant effects of inhibitors of the cyclic AMP-specific PDE IV are markedly enhanced by cyclic GMP elevating agents and by the PDE III inhibitor milrinone. They support the hypothesis that cyclic GMP enhances cyclic AMP-mediated relaxation, possibly through the inhibition of the cyclic GMP-inhibited PDE III.