>>Embolded would be overinterpretation ;-) If base assumption holds true, the file of stacklevel 1 has to stay in memory until every proc/function called from it is done, so basic question should be answered positive re "reasonable". Tempted to write "guaranteed" I thought about options to remove the program at stacklevel 1 - first ideas were on establishing some object "externally", prime contender some other process - either invoked via something asynch, like old DDE or newer object linking and embedding, batch file starting an exe which in turn does activeX getobject() to first vfp - or perhaps just a timer ? There is a good chance even that is not enough by itself to stop vfp from closing itself down after return value of stacklevel 1 is calculated *and* returned - perhaps a memory leak in objects assigned to _screen or _vfp needs to exist as well.
>
>How about something that's evaluated while the exe loads... so main.prg exits but something with a property defined as =(funcCall()) would actually load and do stuff, without main.prg knowing about it. Would that work?

Going deeper into hacking territory - leaving unspoken assumption that the exe built on main loads the runtime ;-)
If you have vfp9 runtimes running the exe or even other compiled vfp program calling the main.prg wit already existing stack (or even debugger in vfp9) might also do the trick...