I have been following Alexa's Traffic Rank for Facebook and Yahoo the past few weeks, and Facebook has finally taken the number 2 spot from Yahoo. This is quite an accomplishment for a relatively young site like Facebook when compared to how long Yahoo has been around. And when compared to all the other services Yahoo provides like news and email.
While Alexa Traffic Rank isn't the most accurate measure of actual traffic it is widely followed and accepted.
The inaccuracy that comes into play with Alexa is traffic is based and measured by the Alexa Toolbar. People who have the Alexa Toolbar installed can see a sites rank according to Alexa in the lower right corner of their browser. Basically each page you visit if you have the toolbar is recorded and sent to Alexa. So people that do not have the Alexa Toolbar are not being counted as visitors by Alexa.
Because of this it really isn't a fair and accurate measurement of actual traffic. For some reason it is widely accept by many webmasters and sites though. Perhaps more along the lines of a rough estimate of traffic.
Consider on a smaller scale two websites that get a true 100 unique visitors per day. One website is a webmaster resource forum that talks about Alexa often, while the other site happens to be a site about sewing. Even though both sites happen to get 100 unique visitors per day, most likely the site about webmaster resources would have a much better Alexa Rank because it's a familiar topic at the site. Many of the visitors and members to the sewing site would probably have no idea what Alexa is, nor would most of them have the toolbar installed.
So in my opinion it's not a very accurate measurement of traffic for most sites. When you get up into millions of visitors per day like Google, Facebook, and Yahoo receive it might be a little more accurate.
Currently the top 5 sites for traffic according to Alexa are: