it is not peer reviewed, *anybody* can stick *anything* there.
I think they've changed things a little bit over time. It does seem like anyone is able to register an account, which would allow them to start submitting papers. But looking at the help pages, I see this on an endorsement system: "Effective January 17, 2004, arXiv.org began requiring some users to be endorsed by another user before submitting their first paper to a category or subject class." They note that this isn't peer review, but it "will verify that arXiv contributors belong [to] the scientific community". They also moderate submissions, and the help page on this topic says: "arXiv reserves the right to reject or reclassify any submission." While also not real peer-review, it "helps to ensure that arXiv content is relevant to current research".
Perhaps some areas are better than others about self-moderating/reviewing submissions. My experience with the astro-ph archive, which I've read for many years, is that most of it is generally good material, often pre-prints of papers that will appear in peer-reviewed journals or conference proceedings. Not all of it is like that of course, but I think there's a lot more signal than noise in the astro-ph section at least. Just my opinion.
Don't get suckered in by the comments -- they can be terribly misleading. Debug only code. -- Dave Storer