To further explore the limitations to maximal O(2) consumption (.VO(2 max)) in exercise-trained skeletal muscle, six cyclists performed graded knee-extensor exercise to maximum work rate (WR(max)) in hypoxia (12% O(2)), hyperoxia (100% O(2)), and hyperoxia + femoral arterial infusion of adenosine (ADO) at 80% WR(max). Arterial and venous blood sampling and thermodilution blood flow measurements allowed the determination of muscle O(2) delivery and O(2) consumption. At WR(max), O(2) delivery rose progressively from hypoxia (1.0 +/- 0.04 l/min) to hyperoxia (1.20 +/- 0.09 l/min) and hyperoxia + ADO (1.33 +/- 0.05 l/min). Leg .VO(2 max) varied with O(2) availability (0.81 +/- 0.05 and 0.97 +/- 0.07 l/min in hypoxia and hyperoxia, respectively) but did not improve with ADO-mediated vasodilation (0.80 +/- 0.09 l/min in hyperoxia + ADO). Although a vasodilatory reserve in the maximally working quadriceps muscle group may have been evidenced by increased leg vascular conductance after ADO infusion beyond that observed in hyperoxia (increased blood flow but no change in blood pressure), we recognize the possibility that the ADO infusion may have provoked vasodilation in nonexercising tissue of this limb. Together, these findings imply that maximally exercising skeletal muscle may maintain some vasodilatory capacity, but the lack of improvement in leg .VO(2 max) with significantly increased O(2) delivery (hyperoxia + ADO), with a degree of uncertainty as to the site of this dilation, suggests an ADO-induced mismatch between O(2) consumption and blood flow in the exercising limb.