On Full Disclosure someone suggested the following:

What you do is have your mail bot send out spam with a twist. As each message is composed and sent, it contains an embedded image of a random name (in fact, it doesn't really exist) that is really a reference number.

For example <img src=http://logging.microsoft.com/verify/123451.jpg border=0 >

Your web server error log will identify every time one of those images was tried and then that gets matched automatically to your database of names.

Now you have:
1. A verified email address
2. An originating IP (can narrow down to what continent they are on or if broadband customers)
3. What OS you are running
4. Possibly what email client or web browser you use.

This is worth big bucks in the form of "email leads" sold by geographic regions and whether they are dialup, cable customers, business, etc.
The only thing I can comment on is that when I had the student forward one of his to me this is what I got (his email has been altered to protect him.. .. )

X-Apparently-To: student9@yahoo.ca via aa.xxx.yy.zz; Wed, 18 Feb 2004 03:18:21 -0800
X-YahooFilteredBulk: 80.57.189.63
Return-Path: <nuhenela@professoremail.com>
Received: from 80.57.189.63 (HELO 66.218.86.253) (80.57.189.63)
by mta203.mail.scd.yahoo.com with SMTP; Wed, 18 Feb 2004 03:18:17 -0800
Received: from 220.14.200.64 by aa.xxx.yy.zz; Wed, 18 Feb 2004 10:17:17 -0100
Content-Length: 0
I had thought at first that it had the information stripped by Yahoo (bulk mail stripping) but I'm beginning to doubt it since he's still receiving it.