It has been hypothesized that the signals of near infrared spectroscopy (NIRS) would reflect muscle O₂ uptake (mVO₂). Although it is not definite that NIRS signals accurately reflect mVO₂, there is every possibility that NIRS signals at least reflect regional O₂ uptake (rVO₂). The phase II kinetics of pulmonary oxygen uptake (pVO₂) is regarded as reflecting mVO₂ at the onset of exercise. To examine whether the rVO₂ on-kinetics measured by NIRS reflects the mVO₂ on-kinetics at the onset of exercise, we compared the rVO₂ as measured by NIRS with the phase II kinetics of pVO₂ at the onset of exercise. Twelve healthy male subjects cycled a Monark ergometer at three different intensities: below the ventilatory threshold (VT) level (below-VT), on the VT level (on-VT), and above the VT level (above-VT), for 6 minutes on three separate occasions. The rVO₂ was calculated from the concentration of oxyhemoglobin and deoxyhemoglobin, as measured by NIRS every 3 seconds. The pVO₂ was determined by the breath-by-breath method. A significant relationship between the amount of increases of pVO₂ and rVO₂ from rest to the end of exercise among all levels of exercise intensity was found (r=0.935, P<0.001). The time constants of rVO₂ (rVO₂-Tc: below-VT: 6.514±2.159 s, on-VT: 7.760±2.035 s, above-VT: 9.532±2.342 s) were significantly faster than the time constants of pVO₂ (pVO₂-Tc: below-VT: 23.8±4.4 s, on-VT: 25.9±5.1 s, above-VT: 26.3±5.7 s) (P<0.001). There was no significant relationship between rVO₂-Tc and pVO₂-Tc for each intensity (P>0.05). We conclude that the rVO₂ on-kinetics measured by NIRS does not necessarily reflect the mVO₂ kinetics at the onset of exercise.