Recommender Systems based on Collaborative Filtering suggest to users items they might like, such as movies, songs, scientific papers, or jokes. Based on the ratings Based on the ratings provided by users about items, they first find users similar to the users receiving the recommendations and then suggest to her items appreciated in past by those like-minded users. However, given the ratable items are many and the ratings provided by each users only a tiny fraction, the step of finding similar users often fails. We propose to replace this step with the use of a trust metric, an algorithm able to propagate trust over the trust network in order to find users that can be trusted by the active user. Items appreciated by these trustworthy users can then be recommended to the active user. An empirical evaluation on a large dataset crawled from Epinions.com shows that Recommender Systems that make use of trust information are the most effective in term of accuracy while preserving a good coverage. This is especially evident on users who provided few ratings, so that trust is able to alleviate the cold start problem and other weaknesses that beset Collaborative Filtering Recommender Systems.