Here is the answer provided by Kevin Struhl in an interesting interview published a few days ago in Current Biology (Current Biology 2008 18:R7):

“I think it is flawed at several levels. Anonymous reviews assume that reviewers are unbiased, objective and without personal or scientific conflicts of interest; this is not always true, especially in competitive situations, and there is no mechanism to detect such problems. Aside from the potential for abuse, anonymous reviews create an inequality between authors and reviewers that is unfair and scientifically unjustified. At many journals, particularly those run by commercial companies as opposed to scientific societies, disagreements between authors and reviewers are often adjudicated by editors with modest scientific accomplishments and experience. I favor a process in which editorial decisions are made by practising scientific experts, reviewers are identified by name, and the signed reviews and author responses published online along with the paper. Lastly, it is unfortunate that the biology community has permitted commercial companies to control most of the journals. Competition among journals and business-related decisions about scientific publishing has seriously distorted the literature, and it has created an artificial rating system that is used to judge decisions about funding and career advancement.”