Online Social Networks (OSNs) have become fundamental parts of our online lives, and their popularity is increasing at a surprising rate every day. Growing popularity of Twitter has attracted the attention of attackers who attempt to manipulate the features provided by Twitter to gain some advantage, such as driving Twitter users to other websites that they post as short URLs (Uniform Resource Locators) in their tweets.Even short URLs are also used to avoid sharing overly long URLs and save limited text space in tweets. Significant numbers of URLs shared in the OSNs are shortened URLs. Despite of its potential benefits from genuine usage, attackers use shortened URLs to hide the malicious URLs, which direct users to malicious pages. Although, OSN service providers and URL shortening services utilize certain detection mechanisms to prevent malicious URLs from being shortened, research has found that they fail to do so effectively.In this project, we developed mechanism to develop a machine learning classifier which detect malicious short URLs. And also shows comparative analysis of detection with various methods.