With the increased use of Internet, Online Social Networks (OSN) has become a part of life for millions of people today. Every day, users of such networks including Facebook, Twitter, etc. execute millions of activities, such as sharing information, posting comments, uploading photos, and updating statuses. The demand on a large amount of information and application that users upload, install, and execute on the social networks makes the social networks an attractive target for attackers. Attackers always misuse human vulnerabilities to launch social engineering attacks. The user behaviors on the OSN make such network begin a fertile area for Malware and attack propagation. Therefore, it is vital to investigate how OSN user behavior affects the vulnerability level of the OSN. In this study, a new model has been built based on Back Propagation Neural Network (BPNN) so as to identify the vulnerability level of the user. This model uses 30 features each of which represents a relation between user vulnerability and attacker policy. One thousand observations for OSN behaviors have been collected by means of surveys in two different countries. The data is used to build training and testing data sets for the BPNN. Performance results show that our model identifies vulnerability level of the user with a high accuracy rate.