I've now started using a very good Naïve Bayesian spam filter, which I've been training for the past week to notice the difference between the nornmal mail I recieve and the spam I recieve.
So far it's become 93.8% accurate, and with training it's getting better.
Out of all the e-mail I've recieved in the past week, 8.88% has been spam.
A word count shows 32.15% of the words I've recieved in my mailbox have in spam!