Robots.txt was my first thought as well, but that is said to not actually block your site from appearing in the results. They'll gather from other sites what the page is about (think <a href=mysite/somepage>how to knit a sweater</a>) and show that as title without page summary. Maybe if it looks like the site is down, they won't bother.

Blocking is based on user agent, they seem to set that reliably and the IP addresses change. You can do some reverse lookup magic but this was way easier than looking up every single IP that visits my site.