Used this extensively for my differential geometry needs (general relativity, gauge theory, etc.) and especially useful for my mathematical gauge theory course I took earlier this year. I did not reach Chapter 7 (de Rham theory) though I read bits of it, and my background in algebraic topology (a course-worth of it) tells me that this chapter is useful as a prototypical example of cohomology theory ---de Rham--- where examples are very easy to grasp in comparison to other (co)homology theories.
If I have to choose, I would say this book is to me better than Lee. It is detailed, easy to understand, and even containing some surprises (like notion of graded algebra, derivations, which are typically glossed over in many courses on differential geometry for good reasons). Some might find earlier chapter a waste of time ("toy differential geometry in Euclidean space") but I think this is very useful especially for people who are not pure mathematicians. A physics student could easily learn pure mathematics from this book thanks to the earlier chapters.
Now, if you skip straight to Section 5 where manifold theory begins proper, this is still a very good book. In fact, I feel that Lee's Smooth Manifolds is a "reference text" because for many purposes (especially introduction) it's overkill. It is very helpful when you want to look for technical details that you might not appreciate when you first learn this (e.g. fibre bundle construction lemma). So as a "textbook", this is the better book for me. Of course, for me I also treat this as a reference text because it actually has enough details for most purposes I need, and only in some rare occasion e.g. proving the manifold construction lemma that I need Lee instead of Tu's text.
In short, if you want to get basic differential geometry right and you only have enough funds to get a book, get Tu.