There’s a great article on spiked science covering the recent stem cell fraud and why peer review ultimately succeeds in keeping research honest.  The scientific tenet that for research to be valid, it must be reproducible by others is pivotal in this.

spiked-science | Article | Peer review and ‘media science’

As we have seen, peer review will not necessarily detect if someone deliberately sets out to falsify data. There is often no way of knowing this until the paper is published and others in the scientific community have the opportunity to scrutinise the work. However, if Hwang Woo-Suk’s paper hadn’t been peer reviewed, and he had gone directly to the media with his ‘results’, it would have taken far longer than six months for the fraud to be discovered and rectified.