Dynamic spectrum method was used to noninvasive measurement of human neutrophilic granulocyte percent for the first time. In vivo measurements were carried out in 21 healthy volunteers, and parital least-squares was used to establish the calibration model of subjects’ neutrophilic granulocyte percent values against dynamic spectrum data. Twenty one samples were classified into calibration set and prediction set, and the calibration was used to establish the calibration model, in which cross validation and leave-one-out method was used to test the best number of factors influencing the PLS calibration models. For calibration set, the correlation coefficient was 0.922, the root mean square error of the calibration set obtained by cross-validation (RMSECV) was1.776%, the biggest relative error was 5.85%, and the average relative error was 4.13%, which promise the good calibration effect. Prediction was carried out to certify the prediction ability of calibration model. And the correlation coefficient of prediction was 0.912, the root mean square error of the prediction set (RMSEP) was 2.930%, the biggest relative error of prediction was 6.74%, and the average relative error was 5.07%, which certify that the calibration model has good prediction ability. Measurement results show that the influences of measuring conditions on spectra can be decreased effectively by dynamic spectrum method and this method can be applied to accurate invasive measurement of human neutrophilic granulocyte percent. It can be a good method for noninvasive blood analysis, which makes great sense to clinical application.