Well, OK, maybe not everything. But perhaps a third of it:

…a third of medical research articles published in major scientific journals and then cited over a thousand times in the literature are later contradicted or have major questions raised over them.

Remember this the next time you hear about a “scientific study,” particularly about politically charged issues, such as global warming. As Iain points out, this is an important point:

We acknowledge that most studies published should be viewed as hypothesis-generating, rather than conclusive.