The idea of using WordNet as a lexical system to improve information retrieval performance is not new. A number of different studies in the past have reported mixed results. In the nineties, several studies reported negative results on the adoption of WorkNet in information retrieval, until 2004, wang and brookes (7) suggested that substantial improvement can be made if the WordNet is used in semantic weigth indexing. Based on this reporting, we developed further the semantic weight indexing by including the semantic distance. The results of the experiments show that contrary to the results reported earlier, the semantic weight indexing actually performed worse than the simple tf*idf weighting. The experiment results more agree with the research done in the early nineties. It confirms that the use of WordNet in information retrieval does not offer much improvement if not any. If it does, the amount of extra preprocessing does not justify the little improvement that if may gain.