Thoreau at High Clearing has a very nice article describing what peer review really is (hint: it ain't auditing). He describes how he handled a paper with the four questions (the details are at his blog, this is just the questions, not the answers)
1) What the hell are they doing here?
2) Do their results make sense? Are the methods solid?
3) Are these calculations relevant to anything? Do they teach us something new?
4) Would these calculations be of widespread interest?