Actually, about 72.2% of scientists are atheists, and 20.8% are agnostics (leaving a scant 7% for believers), though that varies from field to field (more atheists in life sciences, fewer in mathematics). Numerous studies have found that intelligence and religiosity are inversely proportional, and that education and religiosity are also inversely proportional.
http://www.atheists.org/flash.line/atheism1.htm
http://en.wikipedia.org/wiki/Religiosity_and_intelligence
http://library.thinkquest.org/29178/gallup.htm
I'm of the opinion that since most atheists are deconverts (I only have an informal forum poll for that, but out of 200 people, only about a third were raised atheists), it probably has more to do with the process that leads to atheism than atheism itself. The kind of person who is open to the possibility that they are wrong is the same kind of person who I would expect to be fairly intelligent. I think if 80% of the US population were atheists, then we'd see the opposite trend. Atheists who are open to the possibility that they're wrong would be more intelligent, but also more likely to convert if presented with sound reasoning.