news:eeWrMmpnGHA.4772@TK2MSFTNGP03.phx.gbl...

Cryptic though it was, Stefan's suggestion was the correct one -- that is,
redesign my database to avoid using an nvarchar datatype primary key if it
will contain Unicode characters outside the ASCII range.

BUG: In an SQL Server 2000 table connected to an Access 2003 MDB through
ODBC, there can be an erroneous redundant translation between Unicode and
ASCII characters, and the translation only happens on primary key. The
result is that every column in the record will show #Deleted. This issue has
been reported internally at Microsoft as a bug in the ODBC driver and is
awaiting action.

Note that the bug also exists in the ODBC driver in the latest Microsoft SQL
Server Native Client (for Microsoft SQL Server 7.0, 2000 or 2005), Feature
Pack for Microsoft SQL Server 2005 - November 2005, version 9.00.1399:
http://www.microsoft.com/downloads/details.aspx?FamilyID=d09c1d60-a13...

WORKAROUND: However, while waiting for an updated ODBC driver, there is a
workaround if you cannot redesign your database to avoid the nvarchar PK:
1. On Win XP workstation, go to Control Panel, open Regional and Language
Options.
2. On Advanced tab, select Greek in Language for non-Unicode programs.
3. Click Apply button. You may need the setup CD-ROM for Windows XP.
4. Restart the computer. Then all records will correctly show the data.

I expect languages other than Greek will work, but since my problem is
caused by mixed Greek and English in the nvarchar field, I've only tried
Greek. I've seen no other problems when using this workaround, but I have no
non-Unicode applications.

Hope this is helpful to others.

Cheers,

Charax