Abstract: Ash content of coal is a key indicator in coal proximate analysis. The determination of coal ash is made in most cases nowdays by artificial methods. Such methods are all slow process and the determined data can hardly serve as a guide for coal washing process. To tackle this problem, a XRD-based coal ash rapid determination method is developed. As indicated by qualitative analysis of XRD data, the mineral matters assume a spike form on the XRD curve while the organic matters in coal appear as a hump-formed peak featuring a relatively large width; and the ash of coal is positively correlated with the peak-area of mineral matters and negatively correlated with that of the mineral matters. After determination of the peak areas of mineral and organic matters through fitting of XRD curve, two mathematical models for prediction of coal ash values are developed. Study result shows that the model developed based on the two area features a smaller prediction error; and with the use of the model for prediction of the ash values of the coal samples collected from 3 coal preparation plants, the mean values of the absolute errors of the predicted values are 0.7301%, 0.8099% and 0.5015%, respectively. With the use of the XRD-based method, the ash data of dry and ground coal can be obtained within five minutes. Aside its quicker ash determination speed, compared with the radiation method, the XRD-based method has a higher industrial application value for its greater safety.