My mail server receives a lot of spam. My best guess is that on a given day, 70 – 90% of all mail received by my server is spam. In an attempt to reduce the amount of spam my users must sort through, I have been using spamassassin to filter spam on my mail server for several months. I can tell you now that it’s fantastic.

The most recent version of spamassassin (2.55) is a significant improvement over the version I had been running (2.40). The main improvement is the addition of Bayesian analysis of incoming messages. Simply stated, this tool allows spamassassin to “learn” to spot spam. The more it “knows” about what the typical spam message looks like, the better it becomes at stopping it. As a result, I now have to delete just 5 or 6 spam messages a day instead of the 60-80 or more I would be deleting without spamassassin. This is a hit rate approaching 95% – and I get almost 0% false-positives (non-spam messages incorrectly tagged as spam). Outstanding!