I tracked the problem down to the Python urllib.quote() method.

After a little research I found that I should insure any unicode strings are encoded to UTF-8 before passing as a converted ASCII string to urllib.quote(). So I added the following lines to revision 327 of the boto/s3/bucket.py module:

145,146d144
< if isinstance(v, unicode):
<                 v = v.encode('utf-8')

Now my bucket listings no longer break.

I have shared this with Mitch and would hope to see it integrated into the tree soon.