I taught a class at upper undergraduate level (mostly junior students) for a non-majors (mostly engineering students, some economics and political science) at a research university using this book.
It was written in the mid 1980s, and has not seen major updates even though it now comes in 4th edition. One of the assignments I gave was to comment whether you recognize the brand names of the mainframe computers mentioned in the problem. Newer stuff like say the bootstrap or machine learning or anything like that is not mentioned anywhere nearly. True, it has all the major results and techniques, but it was written by statisticians as the first course in statistics for to-be-statisticians, rather than the-only-course-in-statistics-you'll-ever-see for engineers. Thus a lot of things could have been presented in a different way with a different depth of exposure. Say the reliability, arguably a more important topic for engineers than moment generating function techniques, deserves a whole separate chapter, rather than being stuck in a middle of the chapter on continuous distributions. Simulations could have been highlighted throughout the book -- a good fraction of my students would probably be geekier than me with computers. I would unite all the confidence intervals under the umbrella of a single chapter, rather than presenting the CI for mean in one chapter and CI for variance in the next one. And so on. There even were errors in the answers in the end of the book, although you would probably expect the fourth edition not to have any.
Students complained a lot about the book in my class, too. Some said it did not help much, although there were others who did not come much to class (admittedly, I am a pretty boring lecturer) and got B's and A's, so apparently it was of some use to them. The price is of course also an issue: I personally won't pay $120 for book of this quality to sit in my professional library, and it sucks that I have my students buy it.
[Wasserman's [ASIN:0387402721 All of Statistics: A Concise Course in Statistical Inference (Springer Texts in Statistics)]] is a much more modern book, although in all likelihood it would be difficult for my clientelle. My other favorite is Utts' Seeing Through Statistics (with CD-ROM and InfoTrac ), but this one is on the other side of technicality, being too easy. Finally, for engineering students specifically, Ryan's Modern Engineering Statistics (Hardcover) appears to be a much better text, although I have not taught from it, and my recommendation is based on just browsing through the pages and supplementing the current book with examples and problems from Ryan's book.
