As far as I understand, vcards are ascii, however
http://mobilecodes.nokia.com/create.jsp generates vcards
where content is happily UTF8 or latin9 if input so dictates. In these cases, I believe each element should explicitly
define charset.
Risk is, that while in europe phones seem to default translating byte to character using latin9 quite readily,
in other parts of the world, some other charset might
be prioritized first, making vcard saving different content
for different people.

Thanks.