Mark Dvoretsky, Dvoretsky's Endgame Manual, 3rd Edition (Russell Enterprises 2011). 405 pp. $34.95.
I can make this review really quick: if you don't have an earlier edition, buy this one; if you do, then don't.
There's no fundamentally new material here, just some tidying up of the earlier analysis here and there (but where in particular, neither the author nor the publisher gives us any advance clue*) and a slightly improved color scheme for the print. Otherwise, there's no difference between this edition and its predecessor. You might at first be fooled into thinking there are three more pages in the new edition, but that's because the page counted started two pages earlier, with the title page and its overleaf, plus the purely pro forma publisher's preface to this edition.
So while there may not be much reason to get the "mini-upgrade", it's very much worth buying if you don't have an earlier version. This large and widely, rightly praised volume serves a dual function: it provides all the standard theory you'd expect in an endgame textbook, but has a strong practical component as well. Dvoretsky emphasizes typical techniques and mistakes, highlights and focuses in on what's foundational and builds from there, provides numerous exercises, and presents content not only by material but thematic elements as well. The reader isn't just given theory but loads of practical content as well.
If you're around 1800-1900 and up, definitely get the book if you don't already have it. (Even an industrious 1600 could benefit considerably from the book.) There are other good endgame books (and videos) out there, but this is about as close to a must-have book as there is in chess, certainly for endgame play.
* Maybe this silence has been done in the hopes of getting owners of the old editions to buy this one. After all, if they know where the changes are, they can just scan those pages (either figuratively or literally) and not bother buying the new book. Fine, but what about the people who actually buy it - do they really have to sit there with the two books side by side, poring over 400 pages in each, to see what the changes are and if they're significant? Not nice.