We run hundreds of tests on every message that comes through our system. Each test is assigned a numeric value. The message score is the sum of all the tests that were triggered for that message. The higher the score, the more likely the message is spam.
If you look at the message headers, you will see all the tests that were triggered for that message along with their numerical value. For example:
X-Spam-Status: Yes, score=19.509 tag=-1000 tag2=3.5 kill=3.5
tests=[BAYES_95=6, HTML_MESSAGE=0.5, RCVD_IN_BRBL_LASTEXT=1.449,
RCVD_IN_JMF_BL=3, RCVD_IN_MSPIKE_BL=0.01, RCVD_IN_MSPIKE_L3=2.2,
URIBL_ABUSE_SURBL=1.25, URIBL_BLACK=0.1, URIBL_MED=5]
This message scored 19.509 which was higher than the kill level (3.5) so it was marked as spam. The tests that triggered for this message were:
You can adjust the kill level by changing the sensitivity slider on the Domain Settings tab. This page explains the relationship between the sensitivity level and the kill score.