I think it a great book about how to work with old, untested, cryptic or just bad code with the goal of bring it under test. Unfortunately I think the book is two to three times longer than it needs to be. The suggestions that it makes are invaluable, but the examples start to seem repetitive. I also think it might do well to not generalize so much and take advantage of some the tools that are out there today.

In summary I think it’s essential reading if you work with legacy code, but it does lend itself to liberal skimming.

Contains some interesting (and in some cases, novel) ways to deal with legacy code, specifically how to separate dependencies in order to write tests around the code.
For some reason, I wasn’t as drawn in as most people seem to be – possibly because I’m not dealing with legacy code right now. Also, the recipes at the end of the book didn’t all grab me – maybe because a lot of them were described earlier, and maybe because I’d worked out (and used) the mechanics of a number of them already. Still, worth spending your employer’s money on.