Abstract.  Question Answering systems (QA) try to detect snippets of text in a collection of documents, which contain the response to a user’s query. The complexity of QA systems reduces the applicability of these systems to smaller collections of documents. Therefore, QA systems usually employ different tools to reduce to text to work in, such as Information Retrieval (IR) systems. In this paper, we are proposing a Passage Retrieval tool in order to improve the precision and efficiency of a QA system as it was used in the last TREC-10 QA track. Here, we are evaluating this new tool against a standard IR system, and better results have been obtained.