The assumption that cellular oxygen pressure (PO2) is close to zero in maximally exercising muscle is essential for the hypothesis that O-2 transport between blood and mitochondria has a finite conductance that determines maximum O-2 consumption. The unique combination of isolated human quadriceps exercise, direct measures of arterial, femoral venous PO2, and H-1 nuclear magnetic resonance spectroscopy to detect myoglobin desaturation enabled this assumption to be tested in six trained men while breathing room air (normoxic, N) and 12% O-2 (hypoxic, H). Within 20 s of exercise onset partial myoglobin desaturation was evident even at 50% of masimum O-2 consumption, was significantly greater in H than N, and was then constant at an average of 51+/-3% (N) and 60+/-3% (H) throughout the incremental exercise protocol to maximum work rate. Assuming a myoglobin PO2 where 50% of myoglobin binding sites are bound with O-2 of 3.2 mmHg, myoglobin-associated PO2 averaged 3.1+/-.3 (N) and 2.1+/-.2 mmHg (H). At maximal exercise, measurements of arterial PO2 (115+/-4 [N] and 46+/-1 mmHg [H]) and femoral venous PO2 (22+/-1.6 [N] and 17+/-1.3 mmHg [H]) resulted in calculated mean capillary PO2 values of 38+2 (N) and 30+/-2 mmHg (H). Thus, for the first time, large differences in PO2 between blood and intracellular tissue have been demonstrated in intact normal human muscle and are found over a wide range of exercise intensities, These data are consistent with an O-2 diffusion limitation across the 1-5-mu m path-length from red cell to the sarcolemma that plays a role in determining maximal muscle O-2 uptake in normal humans.