The living human skin has complex composition of tissue in several distinctive layers with various thicknesses, and different tissues have different glucose dependence of optical properties. Therefore, the calculation analysis method to find the maximum correlation region containing minimum noise to be used for calibration is worth being concerned in noninvasive blood glucose monitoring. We report an algorithm using correlation analysis to calculate the depth of region, which has the maximum correlation between blood glucose concentration and tissue optical properties. The selected correlation region also has the least noise interference from other skin layers. This method can improve the prediction precision of the noninvasive blood glucose concentration measuring and has a value for application.