I recently followed a link from somewhere to Professor Jeff Erickson's algorithms web site. Erickson, who has been a professor at the University of Illinois for 20 years, maintains a wonderful textbook simply entitled Algorithms.
The biggest advantage of the paper copy, besides that I can read it on my commute where it's easier to read printed material than to bring up my computer, is that it's inexpensive. Most other modern algorithms textbooks nowadays are roughly five times the price of Erickson's book.
The biggest disadvantage of the paper copy is that it is in black-and-white, so the marvelous multi-color diagrams from the online book are rendered in a sort of gray-scale. But the diagrams are carefully-enough built that you can understand them even without the color coding, and if you're really stuck on one you can go online and look at just that one diagram.
Anyway, bottom line: I really like Erickson's book. His writing style is clear and engaging (for a computer science textbook, that is!); he includes a very nice selection of modern, relevant, important algorithms, with plenty of pointers to further areas for the interested reader (many of which are also online at Erickson's site); and the exercises at the end of each chapter are very useful for practicing the techniques that have just been discussed.
Bonus links: if you find Erickson's book hard going because you're a bit out of date on some of the fundamentals, Erickson kindly links to two other great sites: Professor Margaret Fleck's Building Blocks for Theoretical Computer Science, and Lehman, Leighton, and Meyer's Mathematics for Computer Science.
Now, all I need is more time to read.