A method was developed for online monitoring of the constituents of ginsenoside of Folium Ginseng in the column separation and purification process using near-infrared (NIR) spectroscopy technology. Determination method of ginsenoside Rg1, Re and Rb1 was developed by high performance liquid chromatography (HPLC). After collecting 40%-ethanol eluant, their NIR spectra were detected and the contents of Rg1, Re and Rb1 were determined by the above HPLC method. The quantitative analysis models of the above three compounds and the total ginsenosides were established using partial least squares (PLS). During modeling, coefficient of determination (R2) and root mean square errors of cross-validation (RMSECV) were regarded as the indexes to select optimal wave numbers and preprocessing methods. The optimal wave numbers of ginsenoside Rg1, Re, Rb1 and total ginsenosides were all in the range of 12 000.8~7 499.8 cm-1; R2 were 0.988 7, 0.960 3, 0.990 5 and 0.970 1, respectively; RMSECV were 0.059 7, 0.072 2, 0.004 88 and 0.075 5, respectively. A lot of samples, collected during the column separation and purification process of Folium Ginseng extract, were used to validate the predicttion effect of quantitative analysis model of total ginsenosides. As a result, the correlation coefficient of NIR predicted value and HPLC value of total ginsenosides was 0.992 8 and the mean prediction recovery was 100.52%, which indicated that the prediction effect of the developed model was satisfactory. This method was proved to be fast, convenient and precise. It can be used for assaying and quality control of total ginsenosides in manufacture.