My brother has a machine (Gigabyte 7IXE motherboard), that simply reboots, when power off is requested. It does it with APM, or ACPI, so neither one of them is directly at fault. WinXP, of course, works fine. My suspicion is, that Windows simply loops on the power off command, until the machine actually powers off, while OS/2, using APM, or ACPI, simply issues the power off command once, then waits for it to happen. The result appears to be, that the power off doesn't happen, and the power supply sees the "glitch" as being a reset signal, which initiates a power on sequence (reboot <- You can hear something click, at the time that it should turn off). BIOS settings make no difference, so my brother simply waits for the POST, and holds the power off switch, until it does go off. That motherboard will be replaced (it is old), the next time I get to visit him, but I don't believe that the motherboard is defective, I think it may just be a bad design (may be in the BIOS, which could also be used in other motherboards), that the windows wizards have worked around.