"Evaluating Software Architectures: Methods and Case Studies" (Addison-Wesley; ISBN 0-201-70482-X) by Paul Clements, Rick Kazman, and Mark Klein is a book on how to perform software architecture evaluations for the evaluater. Interestingly enough, I've been on both sides of this fence. Generally, I think that I given better evaluations than I have received. I think most people will say that. My advice to add to this book is that if you are in need of an evaluation to first make sure to scope it correctly. Then try to find someone you know to do it. If not, find someone via reference. Finally, if you feel like you and the person doing the evaluation are out of sync, determine if you need to rescope or explain better or you might need to find someone to do the evaluation.
Back to the book, it basically outlines three different methodologies for conducting software architecture evaluations. These are formalizations of what we all have been doing for a long time.
I would recommend this book to anyone who regularly does software architecture evaluations besides it is the first book I've seen that this topic has been discussed in detail. I think it can help improve people's evaluation techniques.