When Internet users are browsing the Web, they always have to wait for the next page to come up after selecting the link to the page, even though the Web server is sometimes idle. To reduce the perceived response time, the server is required to predict users' next HTTP request and pre-process the Web page with remaining CPU resources. This paper introduces various methods to predict users' next HTTP request using classified Web page information, user profiles, and Web logs. Sixteen algorithms and methods are analyzed and compared in this paper. Experimental results show that some methods produce highly probable results with the user of user profiles and classified Web page information.