To gain insight into the role of adenosine (Ado) in exercise hyperemia, we compared forearm vasodilation induced by intra-arterial infusion of three doses of Ado with vasodilation during three workloads of forearm handgrip exercise in 27 human subjects. We measured forearm blood flow (FBF) using Doppler ultrasound and mean arterial pressure (MAP) via brachial artery catheters and calculated forearm vascular conductance (FVC = FBF/MAP) during each infusion dose or workload. We found that about half of the subjects demonstrated robust vasodilator responsiveness to both Ado infusion and exercise, and the other half demonstrated blunted vasodilator responsiveness to Ado infusion compared with exercise. In 15 subjects (identified as "Ado responders"), the change in FVC above baseline was 209 +/- 33, 419 +/- 57, and 603 +/- 75 ml.min(-1).100 mmHg(-1) for the low, medium, and high doses of Ado, respectively, and 221 +/- 35, 413 +/- 54, and 582 +/- 70 ml.min(-1).100 mmHg(-1) for the low, medium, and high exercise workloads, respectively. In the other 12 subjects (identified as "Ado nonresponders"), the change in FVC above baseline was 102 +/- 36, 113 +/- 42, and 151 +/- 54 ml.min(-1).100 mmHg(-1) for the low, medium, and high doses of Ado, respectively (P < 0.05 vs. Ado responders), whereas exercise hyperemia was not different from Ado responders (P > 0.05). Furthermore, infusion of NG-monomethyl-L-arginine (L-NMMA) blunted vasodilator responses to Ado infusion only in Ado responders (P < 0.01 vs. post-L-NMMA) and had no effect on exercise in either group. We also found differences in vasodilator responses to isoproterenol at all doses, but acetylcholine only at one dose, between Ado responders and nonresponders. We conclude that vasodilator responsiveness to Ado exhibits a bimodal distribution among human subjects involving differences in the contribution of nitric oxide to Ado-mediated vasodilation. Finally, our data support the concept that neither Ado nor nitric oxide is obligatory for exercise hyperemia.