EDIT: Thanks to cbehnken for pointing out Doom 3's timedemo problem with dual monitors. That pretty much throws most of the following post in the trash.
Wow. Thanks, PerfeK. I've always noticed that Windows apps that I displayed on the 2nd monitor were much, much more sluggish (WMP visualizations, for example), but it never seemed that having the 2nd monitor enabled made any difference in-game. 90% of the time, my 2nd monitor is showing nothing but the wallpaper (no programs, no start bar, no icons, nothin'), so I figured the performance hit would be minimal. Apparently, that was a false assumption.
I got only a 100pt jump in 3DMark by disabling the 2nd monitor in my display properties, but, as the 2nd poster said, 3DMark doesn't mean much... so, I fired up Doom 3. I ran timedemo demo1 in Doom 3 (at 1600x1200, all high detail) and got 33.4fps with the 2nd monitor enabled and 50.2fps with it disabled! That's not a "slight" difference; that's huge. I'm going to keep the 2nd monitor disabled for the time being and see if benefits other games (particularly D3D games) to a similar extent.
Pulsedrive, you seem to be implying that there's a "proper" way to set up dual monitors so that no performance hit in incurred in games. Would you please elaborate? As far as I can tell, there are only so many nview configurations and they all show the desktop, which would theoretically mean performance hits in all cases.
EDIT: I've tested with a D3D game, RollerCoaster Tycoon 3, and there was absolutely no difference. So, maybe it's something with OpenGL or, specifically, with Doom 3.