There seem to be a few books on the market that are very similar to this one: Nash & Sen, Frankel, etc. This one is at the top of its class, in my opinion, for a couple reasons:
(1) It's written like a math text that covers physics-related material, not a book about mathematics for physicists. I prefer this; you may not. As a consequence, this book is more rigorous than its alternatives, it relies less on physical examples, and it cuts out a lot of lengthy explanation that you may not need. Of course, there are drawbacks to all of these "features" -- you need to decide what you need and what's best for you.
(2) It's most comprehensive, with Frankel coming in second, and Nash & Sen least comprehensive (though they have quite a bit on Fibre bundles and related topics). Nakahara has a chapter on complex manifolds, which is absent from the other two. Nakahara also concludes with a nice intro to string theory, which is absent from the other two as well (though nothing you couldn't find in Polchinski or the like). Actually -- I modify this slightly. Frankel covers less subjects than Nakahara, but with more depth (though also more wordiness -- I quit Frankel about 2/3 through because it wasn't succinct enough and I got tired of it).
Depending on your tastes, I would recommend this book before the other two.
It presupposes that you have an understanding of algebra (groups, rings, fields, etc.) but it has an introduction to the necessary components of topology within. Frankel has presupposes both algebra and topology; Nash & Sen presupposes only algebra.
The first few chapters (homotopy, homology) are rather dry, but the text picks up after that. The manifold chapter is really good, particularly the Lie groups section which gives a geometric viewpoint of the objects which get very little attention in a typical particle physics course. Unfortunately, nothing is said on representation theory, but that can be found in Georgi's book. The cohomology chapter is wonderfully quick and to the point. I found myself having to tell myself to slow down because of the excitement I had in reading it. The Riemannian geometry chapter reads wonderfully and serves as a great reference for all those general relativity formulae you always forget. The end of that chapter has an exquisite little bit on spinors in curved spacetime. The complex geometry chapter is also wonderful. I find myself going back to it all the time when reading Polchinski's string text. The chapters on fiber bundles seem a bit on the overly mathy side, but then again, all the pain is in the definitions which becomes well worth it in the end. I haven't read the last few chapters (spending all of my time in Polchinski!) but I definitely will when I have some spare time. The notation in Nakahara is also really self explanatory and standard. It is written with the physicist in mind who doesn't mind a bit of sloppiness or ambiguity in his notation.
With regards to Frankel, Nakahara is much more modular than Frankel. Each chapter of Nakahara is pretty much self contained whereas Frankel kinda needs to be read straight through. I find it very difficult to just look up a random thing in Frankel and learn about it on the spot, whereas this seems to work in Nakahara just fine. Frankel is a bit more respectful of proper mathematics which also makes it a harder text to read for physicists.
Nakahara is a great text. When I visited Caltech I noticed it on the bookshelf of every theorist that I talked to. Anyone who wants to understand how it is that geometry is so important in modern theoretical physics would do himself a favor in buying this book.