>>
>>Yes.
>>Someone turns off workstation without proper shutdown.
>
>I have often heard the above explanation for the index corruption. But I am wondering, if a table is open by the application but the application is idle, would it still corrupt an index? And why?

Any indexed field may have been updated but the updated DBF and/or CDX (the .cdx file may be updated "later" for all I know) records are still in the OS cache and have not been transferred to the real table.
Keep in mind that that FLUSH without the new VFP9 parameter simply tells the OS that the application is finished with the record. It does NOT ensure that the record is immediately written to the actual table.
Depending on how VFP handles .CDX files it could be even worse for them.

That's one theory.