This is required reading for usability professionals. It's a detailed look at testing, covering everything from test plans and lab construction to data analysis and how to handle unqualified testers who slip through the screening process. It covers lower-budget tests in addition to full-scale ones. It's clear that Dumas and Redish have loads of experience, and they're not shy about sharing it.
The most useful idea I came away with is that testing needs a specific purpose. You can't just test a system's usability; you test, for example, the navigation system or a membership form.
The only flaws here are a couple of omissions. There's no mention of testing web applications, though the principles covered here can be extrapolated to other purposes. And there is no mention of the ultra-cheap guerilla testing tactics. I doubt Dumas and Redish would approve of them, but it would have been very helpful to read about where these alternative methods were effective and not.
This book is a keeper. While my organization will not likely be doing full-blown usability testing for awhile, I'm now more able to evaluate and communicate with third-party testing facilities. And this book will influence all of our other evaluation and assessment testing methods.