>>Yes, that's why I spent over an hour tearing hair out - I checked every which way to Sunday to verify the values but never confirmed until desperate what OLDVAL was returning .NULL.
>
>Don't really know what to suggest - just a couple of SWAGs:
>
>- is it possible your table is corrupted, or somehow actually contains a .NULL. even though it should not be possible?
>
>- has the table record pointer been at EOF() at any time?
>
>- any sort of database trigger etc. in place?

Well I've tried it on 2 different machines and different tables and I get the same result (same .DBC though, copied a few days before).

As regards EOF(), went from USE to APPEND BLANK so my take is no, it has never been to EOF() in the run.

There are no triggers in place.

After this reply I did try to see what was returned if the record pointer was at EOF() and .NULL. is returned there too. Personally I expected an error, but I guess not.

I have gotten around it by comparing EMPTY() entry field with ISNULL of the OLDVAL() return.
But something seems wrong. I surely expected something like this to be documented explicitly if it is the correct return from OLDVAL().

Thanks again
JimN