I occasionally cite web traffic stats from Alexa and Compete, two services that measure traffic across the entire web. It's probably worth pointing out that I have no idea at all whether they're accurate. Compete publishes a monthly count of site visitors based on data from two million U.S. Internet users, so I can compare its numbers directly to the stats I get from Google Analytics. Since the latter is based on actual traffic, it's a reliable metric.
For the Drudge Retort, Google Analytics reports 337,985 U.S. visitors in September and Compete reports 42,815 people for the same period (12.7 percent of Google's total).
On SportsFilter, Google Analytics reports 263,677 visitors and Compete reports 67,906 people (25.8 percent).
On the soon-to-close Cruel.Com, Google Analytics reports 109,334 visitors and Compete 22,028 people (20.1 percent).
I wouldn't expect these numbers to be the same, because every web stats program has different methodology for counting eyeballs. But if Compete was measuring my U.S. traffic accurately, I'd expect the percentages to be close. They're all over the place.