There is a very long list of introductory statistics textbooks, mainly because this is a required subject for many different undergraduate courses. Publishers usually target specific audiences (students in Engineering, Social Sciences, Psychology, Economy, etc.). I have worked with many different introductory stats textbooks and I still keep Montgomery and Runger's volume as my top choice *for engineers* (this is an important remark).
My reasons for this very positive review:
* The authors undertake a good mathematical treatment of most of the topics included, without sacrificing readability and clarity. Engineering students are supposed to have at least some competency in linear algebra and calculus, and this is enough to follow this text.
* The book covers all necessary introductory statistics topics for engineers, including univariate and joint probability distributions, sampling, statistical inference, simple and multiple linear regression as well as ANOVA and experiments with multiple factors. Admitedly, I think that Chapter 6 about sampling and descriptive statistics should have been expanded to include other popular EDA graphs and techniques, but overall I think the book offers a good balance between width and depth of coverage.
* Authors also put strong emphasis in warning against common pitfalls and misunderstandings, not only explaining the failure but also the reasons to understand why that is incorrect. One of my favourite examples is the explanation of why the sample variance must be divided by "n-1" instead of "n", since this point is badly missing in many other introductory texts and students usually get confused about it.
* The content is exemplary well organized. Important terms highlighted, featured frames with relevant definitions, all formulae correctly numbered and referenced. No doubt the authors and editors have spent substantial time in copy-editing. The result is impressive.
* Appendixes include a ton of useful tables and charts for common probability distributions and other systematic calculations. With the growing popularity of statistical software packages this is probably less important in practice than it was 15 years ago, but nevertheless it is an invaluable help for readers interested in following the complete procedures for e.g. inference testing or calculation of confidence intervals.
Finally, I don't understand why Amazon still keeps reviews from past editions (2004 and before) for this up-to-date edition. As for errata, you can check directly with the publishers page that the current list of errors is quite short, mainly including just typos in few solutions, some minor grammar corrections and other insignificant changes. No textbook is free of typos, but I can assure that the up-to-date edition (5th, 2010) of this book is also rock solid in this regard.