In todays world, all activities depend upon the internet. In that Receiving Spam email send messages is a major problem. Many times, this kind of mail contains viruses and hacking links and they affect our system. For solving this kind of problem, we need some method that can filter spam mails and non spam emails. In this paper, we presented one machine learning method that filters spam and non-spam emails. Our algorithm generates the dictionary and features vector and trains them with a machine learning algorithm. Email is one such communication medium that comes to mind when we think of secure communication. As the popularity of email increases, the number of unsolicited data has also increased rapidly. A lot of unwanted stacks of emails called as Spam has created a need for further development Nowadays Machine learning methods have been able to detect and filter out spam emails. The purpose of this current project is to receive a spam email in the morning or effectively using the Multinomial Naïve Bayes method. Naïve approach is a machine-readable algorithm used to classify sample email as spam or not. This filter can be used by other email service providers as fully functional spam filters.