The role of adenosine in exercise-induced human skeletal muscle vasodilatation remains unknown. We therefore evaluated the effect of theophylline-induced adenosine receptor blockade in six subjects and the vasodilator potency of adenosine infused in the femoral artery of seven subjects. During one-legged, knee-extensor exercise at ∼48% of peak power output, intravenous (i.v.) theophylline decreased (P < 0.003) femoral artery blood flow (FaBF) by ∼20%, i.e. from 3.6 ± 0.5 to 2.9 ± 0.5 L min−1, and leg vascular conductance (VC) from 33.4 ± 9.1 to 27.7 ± 8.5 mL min−1 mmHg−1, whereas heart rate (HR), mean arterial pressure (MAP), leg oxygen uptake and lactate release remained unaltered (P = n.s.). Bolus injections of adenosine (2.5 mg) at rest rapidly increased (P < 0.05) FaBF from 0.3 ± 0.03 L min−1 to a 15-fold peak elevation (P < 0.05) at 4.1 ± 0.5 L min−1. Continuous infusion of adenosine at rest and during one-legged exercise at ∼62% of peak power output increased (P < 0.05) FaBF dose-dependently to level off (P = ns) at 8.3 ± 1.0 and 8.2 ± 1.4 L min−1, respectively. One-legged exercise alone increased (P < 0.05) FaBF to 4.7 ± 1.7 L min−1. Leg oxygen uptake was unaltered (P = n.s.) with adenosine infusion during both rest and exercise. The present findings demonstrate that endogenous adenosine controls at least ∼20% of the hyperaemic response to submaximal exercise in skeletal muscle of humans. The results also clearly show that arterial infusion of exogenous adenosine has the potential to evoke a vasodilator response that mimics the increase in blood flow observed in response to exercise.