In the research of massive continuous XPaths evaluation over XML data stream, little has been done on the maintenance of the evaluator when the query set evolves. Due to the massiveness of the XPath set, it takes a high cost to reconstruct the evaluator each time. A new method is proposed to handle this problem. Based on the new transition-rules-counting automata, this method implements the increa se and removal of XPath query dynamically and incrementally, and therefore, avoi ds the high cost of reconstruction. In addition, XML structural constraints DTD can be exploited to improve the efficiency of the incremental maintenance and re duce the possible redundant space cost incurred in the incremental maintenance. The experiments show that the method proposed can efficiently maintain the evalu ator of the continuous query over XML data stream at a low cost of extra space r esource.