It has become a critical issue to effectively retrieve useful information from the Internet as the electronic documents available online has grown drastically fast during the last few years. Most of the currently available information retrieval systems on the Internet are designed to search desired information using keywords. The number of resulting documents provided by these systems are usually more than what is needed since most of them are not highly related, or even irrelevant to the user''''s needs. It creates the so-called "information overload" or "search failure" problems. Due to many of the powerful search engines available, it is more likely to encounter the problem of "information overload". In this study, a keypage-based intelligent meta-search agent is proposed which helps users to easily locate webpages that are more likely fit the users'''' queries. The user may simply provide a web address or an electronic document as the "keypage", the proposed system will then try to locate the candidate webpages and then provide a matching degree for all the candidate pages. Some preliminary results have shown that this system can greatly help users in finding their desired information in a more effective fashion. The proposed system integrates techniques such as Fuzzy theory, SimNet, Parallel Processing and Three-Tier architecture.