Our Anti-Spam service uses a number of clever heuristic tests on mail headers and body text to determine whether messages are spam.  Additionally, our service heavily relies on Bayesian statistical analysis to help sniff out spam messages. If it determines that an e-mail “looks” like spam, it assigns it a positive spam score. The higher the score, the more likely it is that the e-mail is spam.  If the Spam Tag score reaches a certain threshold, customizable by you, the e-mail will be tagged with ****SPAM**** in the subject so that you can decide for yourself whether or not to delete it.  You can set up a filter to place e-mail with this tag in a different folder for later analysis if you wish.

There is also another score you can set called the Spam Drop score.  If the spam score is higher than this number, the system will reject the e-mail outright in order to save our systems from processing what we deem as pure spam.

Please Note: Spam detection is a tricky business.  Spammers are trying every possible strategy to thwart detection.  Therefore, it is given that some non-spam e-mail may be falsely tagged.  So if you observe legitimate messages routinely labeled as spam (false positives), you can add entire domains or specific addresses to your own per-domain white list.  
Conversely, some spam messages may not be tagged (false negatives).  You can add domains or specific addresses to your own per-domain black list.