The receptors through which 2-methylthio ATP (MeSATP), adenosine 5'-O-(2-thiodiphosphate) (ADP beta S), UTP and ATP elicit endothelium-dependent relaxation of noradrenaline-precontracted rings of the rat aorta were characterized by means of a series of antagonists. The acetylcholine-induced relaxation and the degradation of MeSATP, UTP and ATP were also studied. The potency of the nucleotides at producing relaxation decreased in the order MeSATP (EC50 0.24 microM) > ADP beta S (0.43 microM) > UTP (1.09 microM) > ATP (3.53 microM). MeSATP, ADP beta S and UTP did not cause relaxation when the endothelium had been destroyed; high concentrations of ATP still caused some relaxation. The relaxation by MeSATP, ADP beta S and UTP became very small after treatment of the rings with NG-nitro-L-arginine methyl ester; the relaxation by ATP was less affected. Pre-exposure to MeSATP (100 microM) abolished or almost abolished the relaxation normally elicited by MeSATP and ADP beta S, did not change that elicited by UTP and slightly enhanced the relaxation elicited by ATP. Of nine compounds examined as antagonists, six attenuated selectively the effect of some or all of the nucleotides (as compared to acetylcholine): suramin, reactive blue 2, pyridoxalphosphate-6-azophenyl-2',5'-disulphonate (iso-PPADS), pyridoxalphosphate-6-azophenyl-2',4'-disulphonate (PPADS), reactive red 2 and 5,5'-(1,1'-biphenyl-4,4'-diylbisazo)-bis-7-amino -6-hydroxy-naphthalene-1,4-disulphonate (NH05). Decreases of maximal relaxations and slopes different from unity in Schild plots often indicated non-competitive kinetics of the antagonism. For each of the six 'selective' antagonist, the apparent Kd values against MeSATP and against ADP beta S were similar: none of the six differentiated between MeSATP and ADP beta S. Also, for each of four 'selective' antagonists, the apparent Kd values against UTP and against ATP were similar: none of the four differentiated between these two nucleotides (two antagonists did not act against UTP and ATP in the 'selective' concentration range). On the other hand, for five of the six 'selective' antagonists (the exception being NH05), the apparent Kd values against MeSATP and ADP beta S were considerably lower than those against UTP and ATP. At the highest concentrations tested against agonist-evoked relaxations, the antagonists did not alter the removal from the incubation medium, by pieces of rat aorta, of MeSATP, UTP and ATP. It is concluded that nucleotides cause endothelium-dependent relaxation of the rat aorta through two sites: a P2Y-receptor and a P2U-receptor. The receptors may be pharmacologically similar to a bovine endothelial P2Y (P2Y1) and a cloned rat P2U (P2Y2) receptor, respectively. ATP acts mainly through the P2U-receptor. Suramin, reactive blue 2, iso-PPADS, PPADS and reactive red 2 are more potent at the P2Y- than the P2U-receptor. NH05 does not discriminate between the two receptors but is the most potent P2U antagonist so far described.