If you're talking about accepting them as I/O to the mud, that's going to depend on several factors:
The operating system
The mud client
The codebase itself
The codebase is the easiest one to check. In comm.c, for most Diku derivs, you should have the read_from_buffer function which has near the end of it a call to is_print() which will drop any characters from input which aren't printable. This usually drops anything above ASCII 127.
Fixing the OS and the mud client may be out of the question, but they play an important role in whether or not you'll be able to support what you need. What OS and mud client are you using?
I'm actually a little curious about this myself, I've been looking to do something similar to my SMAUG Deriv, Obscurities. I've yet to come across a way to do it yet, though I must admit, I haven't tried all that hard, due to family issues, and other slightly more important projects.
Understand that the Telnet protocol only supports 7-bit ascii, unless Binary or Extended ASCII are negotiated. And as Scandum suggested, many servers strip the 8th bit off, and so do many Mud clients as well. All Telnet clients certainly would.
Edit: I should have said Telnet clients would both strip codes and process certain 8-bit codes as Telnet commands.
like some swedish chars. åäö
cannot figure it out :/