Vista Shutdown Failure Due to Nvidia Drivers

I just fired up a new VISTA system this week and am having a unique problem.

When I press the shutdown button the screen fades to black and shortly there after I loose the video signal to the monitor. When I restart I get an error message that Windows did not shutdown properly.

What's really weird is that if I press RESTART in VISTA the system shuts off completely (no power to fans) and then restarts on it's own just perfectly.

I found that if I delete the Nvidia video drivers the problem goes away and the system shuts down without a hitch. This is not a good solution.

I am using the latest NVidia driver and I have updated my MB to the latest BIOS.

I have pulled other hardware out to see if that makes a diffeerence with no luck.

The MB is a ASUS P5B-E and I have downloaded and installed the latest utilities and drivers from their painfully slow website.

Does anyone have any ideas on this one? I am stumped. Thanks.

System:
C2D E6400
ASUS P5B-E (BIOS 1202 Dated 3/28/07)
MSI Geforce 8800 GTS 640MG (100.65 drivers)
2 Gigs DDR2 PC800
2 160G WD SATA II hard drives running raid 0 on the Intel southbridge
Soundblaster Audigy 2ZS with newest VISTA drivers.

Thanks for resonding.

I looked in the event logs and they show nothing about the shutdown failures. I assume it's because windows is frozen. I have to reset to get the computer going again.

I've found something even weirder. If you click on the arrow next to the power button in Vista and then click on "shutdown" the computer shuts down fine (just like clicking Restart). But pushing the power button still causes the screen to fade to black and freeze.

Define "latest"... as there is the latest 'official' driver which (100.65) released in February, and there is a newer "beta" driver (101.45) (non-certified) that was released in early March.

There are well known 8800 issues with the 100.65 driver in that if you put the PC to sleep, then later wake it up, the 8800 fails to wake, just the PC does. You've not described this, but it does sound perhaps related. This has been fixed in the 101.45 beta driver (or, at least it works for me.)