There is a vast source of information in WWW. How to find the useful information from Internet is an exact issue to be solved. The correct classification of Web pages is the core. Based on the structure characteristics of hypertext, the method of Naive Bayes is adopted in this paper to coordinate the two classifiers that use the text document and hypertext structure. Compared with the two separate classifiers, the combining classifier promotes the correctness of Web pages'classification evidently and steadily.