For on-line book shopping I recommend using Best Book Buys, a book search engine that searches many sites for the best price for your title (it will search new and used book stores). For the best deals, every quarter Labyrinth Books distributes a catalog of various academic titles at substantial discounts (usually well over half off).
I personally recommend against perusing bookstores for good books. If you're going to spend forty hours reading a book, that is a heck of a lot of time that shouldn't be casually left to chance. Instead, find out what books you should read. For fiction, you can get recommendations from a variety of sources, such as friends or on-line lists of recommendations. For philosophy, or other non-fiction, you can ask professors or find what books are being used by courses which are at the right level for you. It's just too easy to read something that is not good philosophy, that is at too high or too low a level, etc.
If you haven't read Brian Leiter's Philosophical Gourmet Report, you should. It ranks philosophy departments, doing a much better job than other rankings, but, more importantly for present purposes, it has a wealth of knowledge that should help in selecting schools. My general advice is to go to the best school that will take you. Though other factors can outweigh this general advice, remember that you will have spent 5-8 years to no end unless you get a job at the end of that time. And better schools are more likely to get you a job (or get you a better job) for a variety of reasons: 1) when hiring, people assess you based in large part on your school; 2) you will likely become a better philosopher by being exposed to better philosophers; 3) you are more likely to get letters of recommendation from better philosophers. So be careful if you're selecting a department just because you're favorite philosopher is there or because they have a cool cognitive science department!
One other fairly important factor, in my eyes, is the package that the school offers. I chose between UCLA and Rutgers largely because Rutgers provided all of their students with a five-year package consisting of three years fellowship (i.e., free ride) and two years as a teaching assistant, while UCLA offered a five-year package consisting of one year fellowship and four years as a teaching assistant. That's two years extra of having to teach while you're trying to write a dissertation. Needless to say, your research will go more quickly if you're not teaching at the same time.
GRE scores, in contrast, are largely determined by your preparation. Devoting substantial time to studying for the tests makes, on average, a significant difference in scores. Study vocabulary (you can buy books for expanding your vocabulary). Consider getting a tutor for the math skills on the test. And above all practice taking many tests so you can learn how to think about the various sorts of questions, so you can learn the math and vocabulary, and so you get used to focusing your mind for the required time (buy the ETS books with actual tests in it; other books create their own tests which are not representative of actual GRE tests). It also makes much sense to take your GRE test early so you can cancel the test if you feel you did poorly, or, better yet, just to practice taking them (i.e. cancel even if you did okay).
At many schools, the writing sample is the most important component. At Rutgers, in any case, every writing sample is read, even if the person's grades and GRE scores are far below par. Unlike grades, but like GRE scores, spending an extra twenty hours on your writing sample can make a big difference in your chance for acceptance. Do not attempt to create a writing sample without help. Ideally you would work with a professor, perhaps starting with your best philosophy paper and, receiving criticisms from the professor, reworking it (i.e. revising it over and over). Even before selecting which paper you'll use, though, you should confer with someone to get advice on the type of paper that is best for a sample.
If you are working with a professor on the writing sample, that can help you to get a letter of recommendation. And you might run the writing sample by a second professor to get additional criticisms and to help get a second letter. Even if you are the shy type, you should force yourself early on to go see professors ocassionally, whether for questions on the material in class or for advice on applying to graduate schools.
Even if you have a good package for your applications, it is easy to make the mistake of applying to too few schools. Try to get a sense of how good the accepted applications are at the schools you're interested in. This is not easy, but often you can just email the graduate director to find out the typical GPA's and GRE scores, which should give you some feel for where you line up (at Rutgers I believe typical GRE scores are around 750 per component). Try to apply to a range of schools, some better and some worse, so you'll have a better chance of getting in at least somewhere. Another common mistake is restricting which schools you apply to based too much upon your current interests. A large percentage of new graduate students come in with one area of interest but end up switching to a different area after a year or two.