Mis-interpreted extended character

Our database ( (PostgreSQL) 7.3.5 ) uses Unicode encoding:

egreek=# \l
List of databases
Name | Owner | Encoding
-------------+----------+----------
egreek | andrew | UNICODE

For some reason, If I try to use an extended character (ASCII code >
127) in a string, I get this peculiar result:

egreek=# SELECT LOWER ( 'Niño' );
egreek'#

(note: it fails to see the closing single-quote character; I hit
CTRL-C to quit the operation)

and:

egreek=# SELECT LOWER ( 'Niño ' );
ERROR: Unicode >= 0x10000 is not supported

(note: I added a space before the closing quote).

It looks to me like the 'ñ' character (spanish n with tilde - ASCII 164,
as far as I can tell) is messing with the parser somehow.

Furthermore -- this is data coming from a form in a web site used
globally, so I have little control over what is passed to me (me =
PostgreSQL DB). I can't (without lotsa hassles) "demand" that
higher-ASCII characters be encoded on the way in, and decoded on the way
out.

Any suggestions/info/hints?

Thanks,

Andrew