The present paper is to study and develop a method for online monitoring of the column separation and purification process of active components that are madecassoside and asiaticoside of Centella asiatica L.Urban using near-infrared (NIR) spectroscopy technology. After collecting 50%-ethanol eluant, we detected their NIR spectra and developed the high performance liquid chromatography (HPLC) assay method of active components. Then, partial least square (PLS) was used to develop linear correlation between their NIR spectra and contents. During modeling, correlation coefficient (R2) and root mean square errors of cross-validation (RMSECV) were regarded as the indexes to select optimal wavenumbers and preprocessing methods. The optimal wavenumbers of madecassoside and asiaticoside were in the range of 12 000.8～7 499.8 cm-1 and 12 000.8～9 750.3 cm-1, respectively; R2 were 96.44 and 96.07, respectively, and RMSECV were 0.084 80 and 0.000 99, respectively. The above developed model was used for online monitoring of the contents of madecassoside and asiaticoside during the column separation and purification process of Centella asiatica L.Urban. The predicted results were satisfactory. This method was proved to be fast, convenient and precise. It can be used in online monitoring and quality control of the manufacturing of madecassoside and asiaticoside.