It is not possible to use ISO 8859-2 keys under XFree86. At present, when you have LC_CTYPE set to any ISO-8859-2 language (cs_CZ for example), it is possible to type only letters (keysyms), which are part of ISO 8859-2 and not ISO 8859-1 (such as ecaron, scaron, ccaron, rcaron etc) but not the characters, which are included in both charsets. This patch allows you to use the characters (keysyms) from the ISO 8859-2 charset, which are also in the upper half of the ISO 8859-1 charset (yacute, aacute, iacute, eacute etc). This is a known bug in X11, and it seems it remains unfixed even in XFree86-3.3.3.1 and RedHat 6.0. :-( We (Czech and Slovak users) have tried to report this to XFree86, Inc., but we've got no response. Maybe RedHat Software will be more successful with pushing the fix up to the original source. The patch is available at http://www.fi.muni.cz/~kas/XFree86-locale.patch The patch is simple one-liner, just add " ISO8859-1:GR" at the end of line 70 (ct_encoding) of /usr/X11R6/lib/X11/locale/iso8859-2/XLC_LOCALE.
Could you also spend two minutes and fix line for "czech" in /usr/X11R6/lib/X11/locale/locale.alias? The alias should be cs_CZ.ISO8859-2, not cs_CS.ISO8859-2. Czechoslovakia (CS) was split to Czech Republic (CZ) and Slovakia (SK) in 1993...
Fixed in errata release XFree86-*-52.