But how does SpamPal know what is spam and what isn't? Well, it uses what are called DNSBL lists. Patterned after the famous MAPS RBL, these are lists of parts of the Internet that in one way or another facilitate spam. Any email you get from a machine on one of these lists has an increased probability of being spam. Some ISPs already block all email from machines on some of these DNSBL lists, but many do not. This is where SpamPal comes in.

You can choose to use any or all of the freely-usable DNSBL lists; SpamPal will look at the machines each email message passed through on its way to your mailbox, and if any of those machines is on one of the DNSBL lists you have chosen then that message will be tagged as spam.

Because not everyone who uses a machine on a DNSBL list will be a spammer, SpamPal has a powerful whitelisting feature that allows you to ignore DNSBL listings for certain senders or for parts of the Internet.

I'd suggest to look at Bayesian plug-in, but don't forget to turn on "learning mode" for the first 4-5 days.