I've come across a reverse lookup site for images that's really neat and very useful. TinEye is still in beta, but it's a googlish beta. The site works just fine.
TinEye works by taking an image provided by you and comparing it to a database of images it has found on the web. It doesn't just look for text around the image or for the metadata; instead it uses matching algorithms to identify what is in the actual image.
I've been using this site fro about a month now, and I have to say I'm impressed. The idea may not be terribly complicated but I'm glad that someone followed through on it. I usually can't use it to find the source (it's lost in the noise), but I've noticed that at least one of the first 20 or so sites that use the image usually credit their source. That's enough to point me in the right direction.
It's been very useful for my photoblog, The Unbound Book.