1. The vasoactive effects of adenosine triphosphate (ATP), adenosine and other purines in the foetal circulation of the human placenta were examined. Single lobules of the placenta were bilaterally perfused in vitro with Krebs buffer (maternal and foetal sides 5 ml min-1 each, 95% O2:5% CO2, 37 degrees C). Changes in foetal vascular tone were assessed by recording perfusion pressure during constant infusion of each purine. To allow recording of the vasodilator effects, submaximal vasoconstriction was induced by concomitant infusion of prostaglandin F2 alpha (0.7-2.0 mumol l-1). 2. ATP (1.0-100 mumol l-1) usually caused concentration-dependent reductions in perfusion pressure. However, biphasic with initial transient increases, or only increases in pressure were sometimes observed. Falls in pressure caused by ATP were significantly reduced by addition to the perfusate of NG-nitro-L-arginine (L-NOARG) (100 mumol l-1) but not NG-nitro-D-arginine (D-NOARG) (100 mumol l-1). They were not influenced by addition of indomethacin (10 mumol l-1) or L-arginine (100 mumol l-1). 3. Adenosine (0.01-1.0 mmol l-1) consistently caused concentration-dependent reductions in perfusion pressure, this effect not being influenced by indomethacin. L-NOARG, but not D-NOARG, reduced the potency of adenosine approximately three fold. L-Arginine, but not D-arginine enhanced its potency by a similar amount. 4. 2-Methylthio-ATP, a selective P2 gamma agonist was approximately 50 times more potent than ATP as a vasodilator agent, always causing decreases in perfusion pressure. 5. Beta-gamma-Methylene ATP, a selective P20 agonist, was approximately 100 times more potent than ATP as a vasoconstrictor, but only caused transient increases in perfusion pressure.6. The rank order of vasodilator potencies of a selection of adenosine receptor agonists was, 2-chloroadenosine>>5-(N-cyclopropyl)-carboxamidoadenosine, >5-N-ethylcarboxamidoadenosine, >2-chloro-N6-cyclopentyladenosine, >CGS-21680 > N6-cyclohexyladenosine = adenosine. Vasodilatation due to adenosine was inhibited by the PI-A2 receptor antagonist 3,7-dimethyl-l-propargylxanthine(DMPX).7. These results suggest that ATP may cause an endothelium-dependent vasodilatation in the foetal vessels of the human placenta via activation of a P2y receptor linked to the formation of nitric oxide(NO). Vasodilatation caused by ATP may mask an accompanying vasoconstrictor effect mediated, via a P2X receptor, in the villous vascular smooth muscle. Adenosine acting on P1-A2 receptors, which are also present in the foetal vasculature, may require synergistic interaction with NO to achieve a maximal vasodilator response.