Arginine vasopressin induces vascular, inotropic and arrhythmogenic effects in the heart. Existing evidence, obtained indirectly, suggests that these effects occur through paracrine endothelial mechanisms. To demonstrate this, vasopressin was confined to the intravascular space by covalent coupling to high molecular weight (2x10(6) Da, vasopresin-dextran) dextran. Isolated guinea pig hearts were infused with equivalent concentrations of vasopressin and vasopressin-dextran. The negative inotropic and coronary vasopressor effects of vasopressin-dextran were similar to those evoked by vasopressin; in both cases effects were reversible. Free dextran had no effect on vascular resistance nor in ventricular developed pressure. The inotropic and vascular effects of both vasopressin and vasopressin-dextran were blocked by the vasopressin receptor antagonist [Adamantaneacetyl(1), o-Et-D-Tyr(2), Val(4), Aminobutyryl(6), Arg(8,9)]vasopressin (Adam-vasopressin), indicating that the effects of the two agonists were vasopressin receptor-mediated. To elucidate possible endothelial intermediaries of these effects, isolated guinea pig hearts were infused simultaneously with vasopressin or vasopressin-dextran and several inhibitors either of synthesis or blockers of receptors of possible endothelial mediators. Only reactive blue 2, a P(2y) purinoceptor antagonist, and suramin, a P(2y) and a P(2x) purinoceptor antagonist, caused a total reversal of vascular and inotropic effects of vasopressin and vasopressin-dextran. Pyridoxalphosphate-6-Azophenyl-2'-4'disulphonic acid, a P(2x) purinoceptor antagonist, was without effect. Our results provide direct evidence that the short-term cardiac effects of vasopressin are due to selective activation of intravascular purinoceptors and suggest that an intermediary of these effects is ATP.