The statistical trainable classification algorithm has unique characteristics. It learns your private correspondence, extracts the distinguishing features from the messages of different classes, and uses that information for analysing the other incoming messages. The algorithm improves its classification quality continuously with each new learned message. It is possible to track back the history of classifying messages and training the plug-in using a filtering log.

The plug-in can delete spam right on the server by headers, using a list of account settings. The accounts for filtering can be added manually or automatically for non-secure connections. The rest of messages are downloaded to email client for filtering by full content, and the messages with high spam ratio are moved to a special folder automatically.

The plug-in supports several complementary methods in addition to statistical algorithm: black and white lists of email addresses, anti-phishing filter, blocking spam by IP address of the sender (using DNSBL services), black and white Regexp-based rules. Several methods are used for filtering graphical spam:
- Blocking spam with inline images, linked to third-party web sites.
- Blocking the messages with animated GIF pictures in attachment.
- Blocking the messages by the type of attachment.
- The statistical classifier is optimized for blocking spam with a picture and some amount of unlinked text.
An additional toolbar is displayed in TheBat interface, providing fast access to the main plug-in functions.