Two excellent and free programs that work using bayesian spam filtering are POPFile and K9.
K9 uses less system resources than POPFile, but I use POPFile simply because you can use it on just about any operating system platform, while K9 is Microsoft Windows only.
Both programs will work with your regular email client provided you get your mail through a POP email server, which the majority of home users do. The program simply sits in between your email client and the server, acting as a proxy that will request your mail, scan it and classify it with a hidden tag that your email program can then be set to delete or put in a special folder depending on your preference.
By using bayesian spam filtering, you will need to train the program what you consider spam and not-spam. The training process usually lasts just under a week, in which you simply mark the email you get as one or the other. In doing so, the program better customizes it to the kind of email you get, hence much more accurate results than the simple and generic rules older spam filters use. On my own personal email, POPFile has a 99.87% accuracy ratio with no false positives.