Rapid and accurate determination of total ash and acid-insoluble ash of Chinese herbal medicine would be valuable for the quality illustration because current approaches are laborious and time-consuming. This study investigated the feasibility of near infrared spectroscopy in predicting simultaneously the total ash and acid-insoluble ash contents of Prunellae Spica (one of the most widely used Chinese herbal medicine). Samples were collected from ten different geographic regions and scanned under the near-infrared spectroscopy region (4000-12000 cm^-1). Models between the spectral data and the results of reference analysis were developed, and subsequently validated. Calibration models based on partial least squares were accurate for the prediction of total ash (R²=0.914; standard errors of prediction=0.373), and acid-insoluble ash (R²=0.905; standard errors of prediction=0.452).