Japanese character encodings are not very well standardized.
There are several distinct industrial standards.

JIS and EUC are the most common, they only use 7 bit characters and 
state-shift sequences.
Shift-JIS (as known as Microsoft JIS) is a variant of JIS that use 
8 bit characters instead of sequences to encode state-shift. It's a
broken scheme, as was to be expected.

All of this does NOT incorporate seamlessly into grander schemes such
as unicode yet...

Nkf is a tool to convert between all the commonest japanese encodings.
It is also able to repair documents which have lost part of their
escape sequences.

The niftiest feature of nkf is that it can guess input encodings, and
usually does this right.