As the spectral pathlength obtained by spectral subtraction approach with different flow blood volumes is uncertain in a near-infrared noninvasive biochemical analysis, it is difficult to obtain a high performance calibration model. Therefore, a pathlength correction space method was proposed to eliminate the uncertain and to improve the performance of calibration. Firstly, the principle of Net Analyte Signal (NAS) and how to obtain a correction space were introduced. Then, a new approach to get NAS was proposed using the pathlength correction space according to the features of spectral subtraction approach. Finally, by taking the intralipid with different glucose concentrations for example, the NAS of glucose was acquired by using the pathlength correction space and a calibration model was established. Compared with the Partial Least Square(PLS) model without pathlength correction space, the correlation coefficient has been improved from 0.978 1 to 0.997 7, the Root Mean Square Error of Calibration (RMSEC) from 77.52 mg/dL down to 25.02 mg/dL, and the Root Mean Square Error of Cross Validation (RMSECV) from 93.01 mg/dL down to 68.22 mg/dL. The analysis result verifies that the pathlength correction space can effectively restrain the influence of different pathlengths and the method can set the stage for the practical application of the spectral subtraction approach with different flow blood volumes.