The data comes from the index released by http://www.dotnetdotcom.org/ (thanks!)

It contains data for 425543 URLs.

The index is dated "200904", and downloaded on 2009-04-24.

They say:

 "A Note On Crawl Quality:
  The downloadable index above contains a uniform sample of pages. We're trying
  to keep this updated as we crawl new pages, but as you can imagine it's a
  little bit tricky :) The data is roughly as fresh as two to three months. The
  oldest pages might be four months old; the newest might be weeks old. Also due
  to our crawling method, Our crawl is probably biased toward English speaking
  sites within the US. Lo siento, mi amigos ;-)"


All URLs had HTTP 200 responses. Content-Type headers (ignoring case and parameters) were:
     424422 text/html
       1227 text/xml
        911 text/plain
         81 text/vnd.wap.wml
         56 binary/octet-stream
        250 (everything else)
(410385 URLs had at least one Content-Type. Some had up to 26.)