My criticism of this is not that it tries to find a biological basis for distressing behaviour -- my criticism is that it is simply bad science, packaged to sell.
Example: Thornhill claims that a study has shown that reproductive-age women are more traumatized by rape than older or younger women who are not in danger of pregnancy, and he takes this as one of the pillars of his argument that rape is a strategy for reproduction.
Problem: His reference for this study is one of his own articles, dated 1991. When you go find this article, you find that it contains a reference to one of his articles, dated 1990. When you find this, it references the study yet again, but in an article of his dated in the early 1980s. When you find this article, you find another reference, but at least this is to the original research -- a study of 27 women that was done in 1974, and which, in fact, the original researchers (not Thornhill) found to indicate that women of all ages and reproductive status were equally traumatized. It was only after Thornhill ran this study through a series of computerized "filters" to factor out things he felt to be extraneous, was he able to turn this interpretation on its head. And it seems that he and Palmer went to extraordinary lengths to make the original data hard to find, in order to obscure the small size and age of the study as well as the original interpretation. The _accepted_ method for citation is to list the original study "as quoted in" one's own article -- not simply to quote one's own series of articles.
He and Palmer consistently make use of obfuscation to avoid answering criticisms of their arguments, as well. For instance, one of his arguments is that scorpionflies regularly "rape" (i.e. force copulation) as a reproductive strategy. At one point, they address the real and relevant question that must be asked -- "How is scorpionfly behavior relevant to *human* behavior" -- by immediately diving off into a rant about the evils of assuming that they advocate biological determinism. In fact, they do not answer the relevancy question at all, but bury it under a load of righteous indignation against a different argument.
Rape of children was not included in any of their analyses; rape of teens and adults was reclassified into only two impossibly broad age categories, thereby obscuring the actual curve of incidents. In fact, the vast majority of rapes target pre-reproductive girls, going by FBI and National Crime Victim Survey statistics -- something that the authors waffle between not mentioning or denying outright, but without naming a source for their statistics. Sex attacks that do not involve a possible result of pregnancy (that is, same-sex rapes, penetration with objects, and penetration of "other orifices", etc.) are not mentioned at all, anywhere in their work -- a strange omission for a book that claims to deal with the entire phenomena of rape. One can only think that they leave these inconvenient facts out of the book because they look awkwardly like evidence against their theory.
This was published as a "popular science" book to make money; such a volatile and emotive topic will always sell. The pity of it is that it would never have passed peer-review as a genuine scientific argument.
I was deeply disappointed, first because of all of the above, and second because the only suggestion they could come up with for actually combating the male tendency to rape was to suggest women dress and speak modestly, and in general do all the things to appear unavailable and spoken for, and to understand that "males are driven by biological needs" -- in other words, the same advice that has failed to work to stop rape in any culture, over a period of centuries.
The book really doesn't advance anything, except Thornhill and Palmer's royalty checks.