First off, the book is very easy to get through - the primary theoretical points are clearly laid out and easy to understand, the selected empirical evidence is interesting, and the style is fluid and coherent. This is the strength of realist theory - clarity of thought, and results in a much more enjoyable read than something by a radical-constructivist or critical theorist. The disagreement over theory is clear from the wide range of ratings in the reviews, but I'd like to briefly cover some of the issues brought up by other reviewers.
Offensive realism, as posited by Mearsheimer is NOT a rehash of Waltz's structural realism but rather adds some important new elements to realist theory. As a result, it is still susceptible to some of the critiques of realist theory in general but also adds new theoretical problems.
Mearsheimer uses Waltz' assumptions on the anarchic nature of the international system and its implication for state behaviour but goes in a very different direction. Using the same assumptions, Waltz believes great powers will essentially be status-quo and defensive while Mearsheimer believes they will be revisionist and aggressive power-maximizers. Mearsheimer thus can avoid the argument against Waltz's defensive realist theory that it leaves no room for transformation of the international system. The potential for conflict is a direct result of the distribution of power in the anarchic system.
The assumptions used by both are by no means "given" and disagreement over them has come from liberal institutionalists, the English School, and the various subsets of constructivist theory. Whether state interests and identities are exogenous or endogenous, and whether there is any room for interests to be shaped by domestic politics, culture, ideology, or institutions is the primary diagreement. Realism says there are only structural variables.
If you think you're a realist, you might want to take note that no prominent realists supported the Iraq War. Realists would also pay no attention to the government of China - a hegemonic democratic China is as dangerous as a hegemonic authoritarian China. Also, the environment and potential for conflict in Europe is the same as in Asia - NATO and the EU, as institutions, are merely tools for great powers to position themselves in hegemonic struggle(this might be included because Mearsheimer said NATO was to be disbanded shortly after the end of the Cold War and needs to find a reason why it is still around). Ideational factors simply have no place in the theory.
Returning to Mearsheimer, he adds another variable to his theory, the impact of geopolitics on state interest. So not only does the distribution of power affect the potential for conflict, geographical factors also play a role. This complicates matters because no longer is there a single variable that can be used to determine causality, and it might be seen as a theoretical crutch. The US and UK both are exceptions to the rule that great powers expand aggressively. Some would say it is because they are liberal democracies, but Mearsheimer argues that it is because they are "offshore balancers", insular states that find it difficult to project power but also derive security from this. An exception to that exception is Japan in WWII, which Mearsheimer explains didn't have a lot of resistance and so couldn't help itself.
Another reviewer wrote of the unverifiability of Mearsheimer's theory. I think that the two variables and the stipulations he puts on power-maximization do make the theory a bit "slippery". Mearsheimer concludes the book by saying that, "Of course, states occasionally ignore the anarchic world in which they operate, choosing instead to pursue strategies that contradict balance-of-power logic." This is a contradiction of the book and realist theory in general. States shouldn't be able to choose how they behave but should be bound by the structure of the system (and geopolitics in Mearsheimer's version) to behave in a certain manner (offensively or defensively according to Mearsheimer or Waltz, respectively).
There are too many exceptions and stipulations to Mearsheimer's rule to make it particularly useful. Whereas the strength of Waltz's theory is its simplicity, Mearsheimer believes he combines theoretical simplicity with policy applicability. He succeeds in neither. Nevertheless, the book is the clearest articulation of offensive realism. Perhaps more comprehensive reading would be Mearsheimer's article on international institutions in International Security Winter 1994/1995, followed by responses from John Ruggie, Robert Keohane, Alexander Wendt, and the Kupchans, and a concluding reply by Mearsheimer.