The final steps of the O2 cascade during exercise depend on the product of the microvascular-to-intramyocyte PO2 difference and muscle O2 diffusing capacity (DmO2). Non-invasive methods to determine DmO2 in humans are currently unavailable. Muscle oxygen uptake (mV̇O2) recovery rate constant (k), measured by near-infrared spectroscopy (NIRS) using intermittent arterial occlusions, is associated with muscle oxidative capacity in vivo. We reasoned that k would be limited by DmO2 when muscle oxygenation is low (kLOW), and hypothesized that: i) k in well-oxygenated muscle (kHIGH) is associated with maximal O2 flux in fiber bundles; and ii) ∆k (kHIGH-kLOW) is associated with capillary density (CD). Vastus lateralis k was measured in 12 participants using NIRS after moderate exercise. The timing and duration of arterial occlusions were manipulated to maintain tissue saturation index (TSI) within a 10% range either below (LOW) or above (HIGH) half-maximal desaturation, assessed during sustained arterial occlusion. Maximal O2 flux in phosphorylating state was 37.7±10.6 pmol·s−1·mg−1 (∼5.8 ml·min−1·100g−1). CD ranged 348 to 586 mm–2. kHIGH was greater than kLOW (3.15±0.45 vs 1.56±0.79 min–1, ptextless0.001). Maximal O2 flux was correlated with kHIGH (r = 0.80, p = 0.002) but not kLOW (r = -0.10, p = 0.755). Δk ranged -0.26 to -2.55 min–1, and correlated with CD (r = -0.68, p = 0.015). mV̇O2 k reflects muscle oxidative capacity only in well-oxygenated muscle. ∆k, the difference in k between well- and poorly-oxygenated muscle, was associated with CD, a mediator of DmO2. Assessment of muscle k and ∆k using NIRS provides a non-invasive window on muscle oxidative and O2 diffusing capacity.