The Mule-UCS package adds or extends support for UnicodeEncoding in Emacs 20 and 21 and XEmacs 21. Get it from
GNU Emacs 21 has limited support for far Eastern script parts of unicode (eg. Japanese, Chinese, Lao). Users of those scripts in utf-8 and similar encodings will almost certainly want Mule-UCS.
XEmacs 21 doesn’t have utf-8 at all by default, Mule-UCS can add it.
Mule-UCS also offers utf-7 as a coding system, if you need to work with that.
In some version of GNU Emacs after 21.3, utf-translate-cjk-mode provides equivalent, and in some ways better, facilities for editing CJK from unicoded files.
Beware that Mule-UCS 0.84 and earlier can corrupt data containing characters it can’t represent, replacing them all with the same character. In contrast, the GNU Emacs 21 UTF-8 coding system maintains unrepresentable characters as the original UTF-8 byte sequences. (Mule-UCS will probably adopt that approach in future.)