MySQL collation for all languages

The accepted answer is wrong (maybe it was right in 2009).

utf8mb4_unicode_ci is the best encoding to use for wide language support.

Reasoning and supporting evidence:

You want to use utf8mb4 rather than utf8 because the latter only supports 3 byte characters, and you want to support 4 byte characters. (ref)

and

You want to use unicode rather than general because the latter never sorted correctly. (ref)

I generally use 8-bit UCS/Unicode transformation format which works perfect for any (well most) languages

utf8_general_ci

http://dev.mysql.com/doc/refman/5.0/en/charset-unicode.html