Twitter, a popular social networking service, enables its users to not only send messages but re-broadcast or retweet a message from another Twitter user to their own followers. Considering the number of times that a message is retweeted across Twitter is a straightforward way to estimate how interesting it is. However, a considerable number of messages in Twitter with high retweet counts are actually mundane posts by celebrities that are of interest to themselves and possibly their followers. In this paper, we leverage retweets as implicit relationships between Twitter users and messages and address the problem of automatically finding messages in Twitter that may be of potential interest to a wide audience by using link analysis methods that look at more than just the sheer number of retweets. Experimental results on real world data demonstrate that the proposed method can achieve better performance than several baseline methods.