Dynamic spectrum (DS) method was used to measure non-invasively human erythrocyte content for the first time. In vivo measurements were carried out to 28 healthy volunteers, and parital least-squares(PLS) regression was used to establish the calibration model of subjects’ erythrocyte content values against DS data. The correlation coefficients were 0.9193 and 0.9058 for calibration set and validation set, respectively. The biggest relative error of validation is 7.96%, and the average relative error is 6.21%. Measurement results show that the influences of individual differences on spectra can be decreased effectively by DS method, and this method can measure accurately human erythrocyte content and it is a promising method for non-invasive blood analysis.