As the number of users connects to the internet increases rapidly, Electronic mail (E-mail) is quickly becoming one of the fastest and most economical forms of communication available, since E-mail is extremely cheep and easy to send. As our lives have become ever increasingly tied up to the online world, the volume of E-mails coming into our inboxes has also been increasing, so the problem of email filtering is a critical one. The current solution usually consists of using E-mail filtering program that can filter incoming E-mail according to user specified rules; this program is the E-mail Filtering Agent (EMFA). EMFA system splits E-mails into categories, this would help to automate the process of sorting through E-mail and applying actions (such as deleting unwanted mail or forwarding or replying messages to a specific address). The agent learns the actions to be performed on E-mail and the features to be used in the classification task from predefined examples in the system. It uses machine learning to classify the messages into two lists: Negative list (that contains unwanted messages) and Positive list (that contains the messages that must be forwarded or replied) and then discard the contents of the negative list and reply or forward the positive list messages. To implement EMFA system, JAVA language was used. It provides a set of abstract classes defining objects that comprise the E-mail system also supporting the creation of sophisticated user interfaces.