The chemical composition is usually acquired by various scientific analyses in Chinese ancient ceramics researches, then they are classified and catalogued by the calculation based on some proper mathematical model, finally the discrimination functions can be got. So the information of provenance and time for the unknown sample can be easily deduced. Rough set theory is a relatively new mathematical tool dealing with vagueness and uncertainty; it can be used to find hidden relations in data and to reveal latent rule of data. Rough set theory has been applied in many areas successfully. In this paper, the basic concepts of rough set theory were introduced, and a process of classifying ancient ceramics was set up primarily. A method to treat the data of the chemical composition of ancient ceramics also was gave, that is, constituting decision tables according to the data at first, then, acquiring a set of classification rules by reduction of condition attributes, reduction of values of condition attributes and reduction of decision rules. All data were analyzed by Energy dispersive X-ray spectrometer. Two examples to apply the method to acquire rules for the classification on the provenance and dating of ancient Chinese porcelain respectively were gave. One example was acquiring different rules for Ding kiln, Xing kiln, Gongxian kiln and Longquanwu kiln, and another was for the blue-white glaze porcelain of Yuan, Ming and Qing dynasties. To verify the rules, all data back were inputed, the results showed that the rate of accuracy of the rules is 100%. It is illustrated that rough set theory could be used in acquiring knowledge about producing area and manufacturing time from composition database of ancient ceramics, and the method could acquire effectively a set of classification rules. In this paper, some other statistical analyses were also compared and their virtues and defects briefly were explained.