>>Any timers around that could switch work area?
>
>Yes, there are timers, but it happens too often just between those lines and nowhere else (in this million+ codeline app).
>I'm beginning to suspect it is a problem that its at the wrong datasession. Still does not explain why it passes the 1st line without a problem.
>
timers should not fire during actual code running, at least if they are vfp timers. C timers might show a more interrupt-like behaviour ;-)
speaking of that: any asynch old OLE embedding ? Was once bitten by the callback disturbing the stack in old code having some privates...
Not sure if bindevented Windows events could show up interrupt-like...