>A while back I posted a problem with an intensive-processing app that slowed to a crawl after an hour or so...I've unearthed the apparent culprit. It's SCAN. After thousands of SCANs, the processing nearly dies. I replaced the SCANs with SQL (even with no optimization available) and FORs, and now the app runs for hours without any slowdown whatsoever.
>
>I found an old SCAN memory leak bug article for FP for MACs, and this sounds the same except it doesn't show up until a great many SCANs are made in vfp5 & 6.

Just a thought, are you doing a SELECT (scanningTable) just before ENDSCAN, if so - it's not needed and slows it down somewhat - may even be source of leak?
(grasping at straws :-)