I ought to say from the outset that I am a regular reader of Ben Goldacre's column in The Guardian, so did not come to this entirely clueless regarding the approach and views of the author. I the think the best way to describe it is Goldacre having a scratch at a few particularly irritating itches. The aims of the book are very good, and to be applauded.
There can be little doubt that this is something of crusade against poorly done medical science, and it is my personal hope that this goes on to have as much social impact as Silent Spring and No Logo. However, don't be deceived into thinking that this covers all of science. Goldacre is a medical doctor and so, with the exception of a couple of chapters on statistics, this is a book on medicine. There is very little mathematics (other than that mentioned above), virtually no physics and just a smattering of chemistry where it overlaps with biology.
One of the slightly uglier aspects of the book is a series of personal attacks around the middle of the book. I had quite a mixed reaction to these. While it does to demonstrate some points that were much needed, the near unrelenting attacks does show a certain insensitivity on the part of the author. Here, we have to consider how well the author communicates; I have heard Goldacre talk and he tends to come across as quite reasonable, with a hint of mild exasperation, but in the book he comes across as a bit more patronising and sneery. This is most evident in the introduction where he states that anyone who dares disagrees with him is wrong; whereas a true scientist would always be open to the possibility that they themselves may be wrong. His choice of target is also interesting, as Gillian McKeith does not appear to have caused any physical harm by her lack of understanding, his later target Matthias Rath, almost certainly has. In this respect, even though the chapter on McKeith is one of the most famous, it is also one of the most unnecessary.
The last third of the book looks at one area in which Goldacre himself works: the media. Mentioned in the introduction, he goes on to look at CP Scott's “two worlds” hypothesis between the humanities and the sciences, with an exasperated moan that the media is dominated by humanities graduates who cannot understand the sciences, before looking at some of the consequences that has had.
Overall, the aim of the book is noble and many of the examples are well-put. Given that Goldacre advocates (rightly, I believe) transparency in science, it would have been nicer to have included more references at the back since at present they have the appearance of being cherry-picked. For anyone interested in medicine this is an essential read, though I think the focus on medicine does make the title a bit misleading. If the statistics chapters interest you, then I would recommend Leonard Mlodinow's The Drunkard's Walk as a follow-up which expands in more detail some of the topics touched upon by Goldacre.