Near-infrared spectroscopy (NIRS) is a well-known method used to measure muscle oxygenation and hemodynamics in vivo. The application of arterial occlusions allows for the assessment of muscle oxygen consumption (mV̇o₂) using NIRS. The aim of this study was to measure skeletal muscle mitochondrial capacity using blood volume-corrected NIRS signals that represent oxygenated hemoglobin/myoglobin (O₂Hb) and deoxygenated hemoglobin/myoglobin (HHb). We also assessed the reliability and reproducibility of NIRS measurements of resting oxygen consumption and mitochondrial capacity. Twenty-four subjects, including four with chronic spinal cord injury, were tested using either the vastus lateralis or gastrocnemius muscles. Ten healthy, able-bodied subjects were tested on two occasions within a period of 7 days to assess the reliability and reproducibility. NIRS signals were corrected for blood volume changes using three different methods. Resting oxygen consumption had a mean coefficient of variation (CV) of 2.4% (range 1–32%). The recovery of oxygen consumption (mV̇o₂) after electrical stimulation at 4 Hz was fit to an exponential curve, which represents mitochondrial capacity. The time constant for the recovery of mV̇o₂ was reproducible with a mean CV of 10% (range 1–22%) only when correcting for blood volume changes. We also examined the effects of adipose tissue thickness on measurements of mV̇o₂. We found the mV̇o₂ measurements using absolute units to be influenced by adipose tissue thickness (ATT), and this relationship was removed when an ischemic calibration was performed, supporting its use to compare mV̇o₂ between individuals of varying ATT. In conclusion, in vivo oxidative capacity can be assessed using blood volume-corrected NIRS signals with a high degree of reliability and reproducibility.