SPAM was originally abbreviated for ”Self Promotional Advertising Message”, but now the most popular definition characters as ”unsolicited bulk mail”. [1]. Email spam especially is the major problems of the today Internet, bringing an increase in network bandwidth, and financial damage to individual users. Among the approaches developed to stop spam, filtering is an important and popular one. In this project we give several state-of-the-art of different ways of machine learning approaches, and also a brief description of other branches of anti-spam protection.