Abstract: This paper studied a rapid coal ash determination method based on the near-infrared diffuse reflectance spectral analysis by build mathematical regression model between the near-infrared diffuse reflectance spectra of coal samples and the ash.The determination is based on the partial least squares regre ssion(PLS)algorithm,while different spectral pre-processing methods including smoothing,Multiplicative Scatter Correction(MSC) and derivation were applied to improve the prediction ability of the mathematical model.The ash level of real coal samples was experimental estimated,which indicated that the determination results were very well.The method can get high precision and stability which leads to a good potential application in industries.