>I don't think VFP "moves" the record pointer. After tableupdate() is successful VFP assigns a new number to the RECNO() but nothing physically is moved. But in my case, of tablerevert(), it seems like VFP actually moves the pointer to a new (phantom) position.

I'm pretty sure the pointer is moved (if we're talking about table buffering). In your case since the newly added record had to be reverted, the pointer positioned on the EOF(). I see nothing wrong here.