Question answering systems (QAS)s aim to directly return precise answers to natural language questions. Retrieving and re-ranking passages are viewed as the most challenging tasks in a typical QAS and still require nontrivial effort. In this paper, we propose a novel approach for retrieving and reranking passages using n-grams and SVM. Our n-gram based passage retrieval engine relies on a new measure of similarity between a passage and a question. The retrieved passages are further re-ranked using a Ranking SVM model combining different text similarity measures in order to return the most relevant passage to a given question. Our experiments and results have shown promise and demonstrated that our approach is competitive.