I don't know where you're going with that. Not saying it's not possible. What if - instead of the bsod, you were brought to the equivalent of a WinDBG output of the crash? You could diagnose the cause of the crash while everything was halted, figure out what's in a corrupted state and what isn't - unload modules, stop threads, change memory values, do whatever you want to do. Then if you felt comfortable enough, you could allow the operating system to "resume". Look, I'm not saying that this is an end-user thing to do, and lots of time you'd probably be better off letting the machine crash. But if I had that kind of power in the halted state - I could even find my unsaved documents in memory and store them in some way without even allowing the OS to continue. And I wouldn't just have the backtrace available to me, I'd have the entire contents of memory such that I could !process 0 7 and look at every thread in the system. Hey, I could be wrong - since this kind of power doesn't exist - but like I said, I have been able to reanimate a crashed dev studio by messing around while it was in it's halted state. I'd think it'd be possible to do the same in the kernel in certain cases. Are you saying with certainty that there is never a case where this is possible? it is true though, that lots of corruption could have happened before some trap actually caught the bad state and KeBugChecked. Maybe you're right. I just don't think that you are. With all due respect.