>>I read the parallel thread but I'm not convinced of this - if VFPOLEDB was going to choke on reserved words I'd expect it to happen everytime not just intermittently.
>
>Exactly, this is why I didn't celebrate so far. To me, this situation is not making any sense. But, then, I tried again to look and look again in the code and I really can't find anything else to adjust.

There's still the option I suggested a while ago - give each connection a GUID when it's created and look for cases where the same connection is used by more than one hit?

Regards,
Viv