You just received an unpleasant message and you want to ill-treat the author.
Or an individual spits unbearable opinions on a NG and you want to kill his dog.
You look at the headers and both comments and message-IDs point the origin to a remailer.
What can you do?
Cool off:
In most so-called civilized countries, you are not allowed to exact justice yourself.
Take a fresh look and check the means you can have enacted.

Your writings raised hell in numerous NG, and quite a few hot brains are preparing tar and feathers.
Are you well-protected against any unwanted consequence to your declarations?
It would have been wise to check before.

An anonymous remailer
    uses one or many inbound messages
    to build one or many outbound messages
    and does not *create* nor *keep* any evidence of the cross-reference Inbound-Oubound.
Hence it is impossible to track back to the author of the messages, using the sole message-ID on arrival:
    the trail stops immediately.

Let us start with an example:
A unique emitter A sends a mail-bomb of  1000 identical messages, all 1000k (after encryption)
  to a single recipient B through the remailer network, all using the same path.
Inside the network, messages will be remixed (split into 28k chunks) resequenced....
  but in the very end 1000 messages (- transmission loss )  each 1000k (-25% at each remailer hop)
will land at recipient B.
Do you really believe that 'somebody' able to watch both mail outbound from A and inboud to B
  might have the *slightest* doubt about what is going on?
And all that without having to know anything about which path has been used,
leave alone having to decypher or even track intermediate messages !!!
Statistic traffic analysis allows to solve problems much more complex !!!

Do not use remailers for direct mail-bombing:
  not only it is not 'acceptable use'
  but you might get caught without a single remailer being compromised
Same for mass expeditions, or periodical, or abnormal size....
   change path, use latency, remix, generate dummy traffic....

If a sender broadcasts the same information under various screen names, and the content is identical
    by the word !!!!
    or with identical ideas
    or with an identical way to formulate ideas
    or with identical misspellings, faulty grammar, or punctuation....
It will be enough if one of the screen 'gets caught'  (i.e. points to the real name)
    to have the whole lot of messages be attributed to him.
If you got multiple personnalities, better for you they are well split !!!

Most documents from Microsoft (Excel / Word / VB ....) incorporate an identifier to the resulting code:
your user name or your licence number and/or....
Other software editors do the same.
Color copiers have a built-in identifier in their canvas, out of sight
.....
Avoid attachements of anything different from plaintext....

Let us admit person X is suspected of serious crimes.
A judge rules his telephone to be tapped, his outbound encrypted mail to be intercepted.
A judge rules the remailer admin to decypher a message.
  will the admin refuse?
  and so on to the final recipient
Or you choose a remailer willing to refuse. (Cf: Privacy)
Or you use a chain of remailers sitting in various jurisdictions,
       and which have a policy of periodical destruction for keys.

Unless there is very strong suspicion against them,
and their actions are serious enough to trigger immediate cooperation from numerous transnational jurisdictions
a carefully-thought use of remailers and simple security rules should ensure total security for end-users...
At any rate,
    nor enquiries performed by individuals
    nor prosecution for political opinion under one particuliar jurisdiction
pose a real threat, except for gross misuse of remailers...