Spam has serious negative on the usability ofemail and network resources. Spam is flooding theinternet with many copies of the same message, in anattempt to force the message on people who would nototherwise choose to receive it. And despite the evolutionof anti spam software, such as spam filters and spamblockers, the negative effects of spam are still being feltby individuals and businesses alike. To prevent thisadvance techniques are necessary. Our proposed methoddivides e-mails in spam class and non spam classaccording to different attribute values of spam. analternative approach using a neural network (NN)classifier brained on a corpus of e-mail messages fromseveral users. The features selection used in this work isone of the major improvements.Keywords:-Spam, E-mail classification, Machine learningalgorithms ,Emails classification, Document similarity , Documentclassification, Feature extraction, Subject classification, Contentclassification.