Maximal O2 delivery and O2 uptake (VO2) per 100 g of active muscle mass are far greater during knee extensor (KE) than during cycle exercise: 73 and 60 ml. min-1. 100 g-1 (2.4 kg of muscle) (R. S. Richardson, D. R. Knight, D. C. Poole, S. S. Kurdak, M. C. Hogan, B. Grassi, and P. D. Wagner. Am. J. Physiol. 268 (Heart Circ. Physiol. 37): H1453-H1461, 1995) and 28 and 25 ml. min-1. 100 g-1 (7.5 kg of muscle) (D. R. Knight, W. Schaffartzik, H. J. Guy, R. Predilleto, M. C. Hogan, and P. D. Wagner. J. Appl. Physiol. 75: 2586-2593, 1993), respectively. Although this is evidence of muscle O2 supply dependence in itself, it raises the following question: With such high O2 delivery in KE, are the quadriceps still O2 supply dependent at maximal exercise? To answer this question, seven trained subjects performed maximum KE exercise in hypoxia [0.12 inspired O2 fraction (FIO2)], normoxia (0.21 FIO2), and hyperoxia (1.0 FIO2) in a balanced order. The protocol (after warm-up) was a square wave to a previously determined maximum work rate followed by incremental stages to ensure that a true maximum was achieved under each condition. Direct measures of arterial and venous blood O2 concentration in combination with a thermodilution blood flow technique allowed the determination of O2 delivery and muscle VO2. Maximal O2 delivery increased with inspired O2: 1.3 +/- 0.1, 1.6 +/- 0.2, and 1.9 +/- 0.2 l/min at 0.12, 0.21, and 1.0 FIO2, respectively (P < 0.05). Maximal work rate was affected by variations in inspired O2 (-25 and +14% at 0.12 and 1.0 FIO2, respectively, compared with normoxia, P < 0.05) as was maximal VO2 (VO2 max): 1.04 +/- 0.13, 1. 24 +/- 0.16, and 1.45 +/- 0.19 l/min at 0.12, 0.21, and 1.0 FIO2, respectively (P < 0.05). Calculated mean capillary PO2 also varied with FIO2 (28.3 +/- 1.0, 34.8 +/- 2.0, and 40.7 +/- 1.9 Torr at 0.12, 0.21, and 1.0 FIO2, respectively, P < 0.05) and was proportionally related to changes in VO2 max, supporting our previous finding that a decrease in O2 supply will proportionately decrease muscle VO2 max. As even in the isolated quadriceps (where normoxic O2 delivery is the highest recorded in humans) an increase in O2 supply by hyperoxia allows the achievement of a greater VO2 max, we conclude that, in normoxic conditions of isolated KE exercise, KE VO2 max in trained subjects is not limited by mitochondrial metabolic rate but, rather, by O2 supply.