The documented low levels of reliability of the peer review process present a serious challenge to editors who must often base their publication decisions on conflicting referee recommendations. The purpose of this article is to discuss this process and examine ways to produce a more reliable and useful peer review system.