Terry,

I know I can change it to Select first and then loop and delete one by one; and it does work faster that way.
However, the question was, what makes it work slow in this particular scenario:
1. It is networked table I delete from
2. It is opened at least one more time by another VFP instance
3. I do not issue FLOCK on it prior to DELETE
4. It has nested SELECT or in some other way requires joining of the table being updated with an intermediate cursor in VFP 9

Because it works fast as it is if
1. It is local table to my machine
2. I do FLOCK first or ensure that it is not opened by anybody else
3. The command dos not involve nested Select

I need to understand what syntax to avoid without running into these performance issues later. And I prefer using standard SQL because of portability issues.

--Andrew