I know there have been posts here on this, but I can't seem to find a solution.  My bayesian filter has been on now for quite some time (over 30,000 messages sent and received), but if I pull spam from the quarantine to test the bayesian probability, everything is "0".  I also noted in another post that:

SELECT     tblQuarantine.RejectID, tblRejectCodes.RejectDesc, COUNT(tblQuarantine.RejectID) AS Total
FROM         tblQuarantine INNER JOIN
                      tblRejectCodes ON tblQuarantine.RejectID = tblRejectCodes.RejectID
GROUP BY tblQuarantine.RejectID, tblRejectCodes.RejectDesc

Should be run on the database.  I performed this query, but there is no reject codes showing up for the bayesian filter.  Shoudn't spam that has already been identified show me some percentage of probability when run against the bayesian filter, or am I confused (a REAL possibility). 

    Thank you!

I seem to be having the issue here. I have turned the "Threshold" down to 85% probability, and nothing seems to get caught. I have to admit I am not sure how large an issue that is for me, we are already blocking some 70% of email coming in, but all the blocks are for attachment types, and the few keywords I block.

Any email I get appears to be blatant spam, when tested, comes back at 0% probability of being SPAM.  I have been lowering the probability to see if I can catch anything, but to no avail. I have tested at as low as 30% probability.

Roberto it might be helpful to post the results of your inspection of their zip files. This might reduce the number of questions you receive about this topic.

Lee