Abstract: The main error factors on near-infrared spectroscopy (NIRS) measurements for deep tissue are (1) thickness of overlying tissues, (2) oxygenation of surface tissues, (3) difference between an assumed value and an actual value of scattering coefficient for tissues, and (4) concentrations of hemoglobin derivatives. These factors affect the accuracy on time-resolved NIRS, intensity-modulated NIRS, spatially resolved NIRS, and continuous wave NIRS. In this paper, we present the accurate oxygenation measurement in surface and deep tissues using continuous wave and spatially resolved NIRS. The relationship between spatial distribution of light intensity and absorption coefficient of the muscle was obtained from the Monte Carlo simulation. Fat thickness greatly affected the absolute value of hemoglobin concentration. The influence of skin oxygenation caused several percent error factor of muscle oxygenation. We obtained curves to correct the influences. To reduce the error due to assumed scattering coefficient, a simple method for determining optical properties with laser rangefinder was practical and useful. Moreover, the absolute values of hemoglobin derivative concentrations were successfully measured by spatially resolved NIRS. It was also helpful for accurate oxygenation measurement. The combination of these measurements and corrections would be essential for an accurate NIRS.