1. The purine nucleotide adenosine-5'-triphosphate (ATP) exerts pronounced effects on the cardiovascular system. The mechanism of action of the vasodilator response to ATP in humans has not been elucidated yet. The proposed endothelium-derived relaxing factors (EDRFs) were studied in a series of experiments, using the perfused forearm technique. 2. Adenosine 5'-triphosphate (0.2, 0.6, 6 and 20 nmol dl(-1) forearm volume min(-1)) evoked a dose-dependent forearm vasodilator response, which could not be inhibited by separate infusion of the nonselective COX inhibitor indomethacin (5 microg dl(-1) min(-1), n=10), the blocker of Na(+)/K(+)-ATPase ouabain (0.2 microg dl(-1) min(-1), n=8), the blocker of K(Ca) channels tetraethylammonium chloride (TEA, 0.1 microg dl(-1) min(-1), n=10), nor by the K(ATP)-channel blocker glibenclamide (2 microg dl(-1) min(-1), n=10). All blockers, except glibenclamide, caused a significant increase in baseline vascular tone. The obtained results might be due to compensatory actions of unblocked EDRFs. Combined infusion of TEA, indomethacin and l-NMMA (n=6) significantly increased the baseline forearm vascular resistance. The ATP-induced relative decreases in forearm vascular resistance were 48+/-5, 67+/-3, 88+/-2, and 92+/-2% in the absence and 23+/-7, 62+/-4, 89+/-2, and 93+/-1% in the presence of the combination of TEA, indomethacin and l-NMMA (P<0.05, repeated-measures ANOVA, n=6). A similar inhibition was obtained for sodium nitroprusside (SNP, P<0.05 repeated-measures ANOVA, n=6), indicating a nonspecific interaction due to the blocker-induced vasoconstriction. 3. ATP-induced vasodilation in the human forearm cannot be inhibited by separate infusion of indomethacin, ouabain, glibenclamide or TEA, or by a combined infusion of TEA, indomethacin, and l-NMMA. Endothelium-independent mechanisms and involvement of unblocked EDRFs, such as CO, might play a role, and call for further studies.