Near infrared spectroscopy (NIRS) is now a commonly accepted method for the measurement of oxidative metabolism of brain and muscle tissues. Nevertheless, this technology suffers from the error caused by the homogeneous single layer assumption in the calculations of concentration changes of light absorber chromophores using either diffusion theory or modified Beer-Lambert law. Underestimation of muscle oxidative metabolism for muscles having thicker fat layer above is a particular case. Due to this uncertainty. statistical analysis can be problematic in reviewing the results across subjects having different fat thicknesses for muscle studies. In this study, partial pathlength method with two detectors based on modified Beer-Lambert law extended for heterogeneous medium with homogeneous layered reo,ions is investigated. Using Monte Carlo simulations. comparison between this technique and single homogeneous layer assumption is done. Optical coefficients of fat and muscle layers are chosen typical for muscle tissue measurements. In the simulations, change of absorption coefficient in muscle layer was made much bigger than in fat layer. It has been found that for 2-detector partial pathlength based method, fat and muscle layer absorption coefficient change estimates are better than the homogeneous medium based modified Beer-Lambert law estimates in all simulated cases.