This is one of those topics that bears mentioning periodically - screen resolution statistics are misleading.
Often when people are thinking of creating a fixed-width layout, they are doing so due to what their statistics are telling them. If statistics are saying that 99.5% of their views are from vistors browsing on high resolutions, many people feel safe with breaking out of the box and going above the fixed width recommendation of designing for a minimum of an 800x600 screen resolution. Unfortunately, those statistics are misleading.
To explain why, I'll use myself as a common example of a misleading visitor. The resolution that would read from a visit from me would be 1400x1050. However, that is not the amount of space I actually use when I'm browsing. I like to keep space around my windows so that I can stack things and watch them while I'm working on something else. Like this screen shot I took of my 1400x1050 resolution desktop (image is resized to 800x600 so as to avoid being ridiculously large for viewers who don't use the resolution I do ;)):
You see why it would be misleading to assume that just because your stats say I am using 1400x1050 -- that I actually am? The more screen real estate that a person has, the more prone they are to stagger windows just like this. Additionally, take a look at how small my viewport (the portion where your website is actually show, not including the tool bars) is in comparison to my browser window. So... keep that in mind the next time you are tempted to start making a fixed-width webpage just a little bit larger because your stats 'claim' that your users all have loads of screen space.
Cross-posted to take your comments on Beyond Caffeine.