Biometrics-based hand authentication is among the most popular biometrics used to automatically characterize a person especially in forensic applications. Hand recognition systems are able to confirm or deny the identity of a claimed person because they do not cause anxiety for the users. However, different individuals may have almost similar hands. Therefore, the performance of the hand verification process depends highly on the hand descriptors. In this paper, we propose a new approach for personal verification based on Scale Invariant Feature Transform (SIFT). This transform proved its high distinction and efficiency in many applications especially in object recognition and video tracking. Two public databases have been used to evaluate performances. Experimental results show promising recognition rates by achieving 94% for IITD hand database and 98% for Bosphorus hand database.