Lessons Learned in Software Testing by Cem Kaner, James Bach, Bret Pettichord (2002)

I like it.
It's a long-play book, that you can (and want) to read from time to time. It is designed for coming back: it has 293 lessons with very clear topic and with some thoughts about this topic. So, if you want to read something about specific problem, then it is quite easy to find thoughts about it.

I like that authors don't afraid of short explanations - for example, lesson 4 have only 5 rows and it's nice, because there is no unnecessary information in it. There is a lot of big books with big chapters where concentration of useful information is actually very low. Not in this book.
I am agree with Tim Lister, who said (on page XVIII of this book): "If you have never participated in a serious software effort, this book will be too heady for you." This book is for experienced testers, not for juniors. It can be useful for those who have already encountered with some problems in testing and have been already thought about them.

But I am not quite agree with further "rule of using" this book (also from Tim Lister on same page): "Don't drink the entire bottle" Meaning "don't read this book in one sitting". I've read it in one sitting for the first time, which gave me general overview of it's content. For me it is not very important to understand all things in first time, it is more important to know what sort of information you can find there so you can return to it later.

So I think every tester should have their own copy of "Lessons Learned in Software Testing" - it is not that kind of book that you can borrow. It is very useful when you can come back to it whenever you want.