So this was a post I made to HoboNinja last year; although the situation's a bit different, I hope it'll help. Although I don't have a PhD, I am a (happy) second-year graduate student at UMich.
For starters,
http://talk.collegeconfidential.com/graduate-school and
http://forum.thegradcafe.com/. The 'Grad School Admissions 101' thread on College Confidential is a big help, and you can find threads for previous applicants (generally in the biomedical sciences, but there's sub-forums for your intended major) detailing stats and such.
Second, don't be paranoid about the 'stats inflation' on that site: graduate admissions is a holistic process, more or less. Keep your GPA, especially in your major, as high as possible (most programs have an average GPA of ~3.5 or so), and study for the GRE a few months before you take it. From what I've heard, GREs are useful in applying for funding, so make sure you do as well on that as you possibly can. It's not too difficult, though; I got my score after about three weeks of studying, and it's more about 'knowing how the questions are asked' than 'knowing the answers'. You'll see when you start studying.
Third, get in on some research as soon as possible. Connect with your professors, ask questions, and--if you happen to like one of your physics professors--ask what research they're doing, and if they could use a hand. You won't get paid for it (unless you're hired on as a lab tech, but most undergrads won't be), but you will get some independent research credit out of it (which can help boost your GPA), and--also importantly--you network yourself with professors.
This will be invaluable when it comes time to choose who will be writing your letters of recommendation, because you'll have people who will be able to say far more than 'so-and-so took my course, got an A, and was a good student.' I felt confident asking one of my cell/molec teachers to write me a letter because I spoke with him so much over the three or four classes I took with him. He was able to write me a strong letter, which leads me to point four.
Point four: research experience and letters of recommendation generally carry the most weight at graduate programs. When I went to the PREVIEW weekend at Michigan, we were given a powerpoint presentation about what Michigan looks for in its strongest applicants, and bolded at the top were research experience and letters. More than simply 'doing' the research, try to understand WHY you're doing it. Try to understand the experiment and its design. It behooves you both since, if you want to go to graduate school, you'll need to anyway, and has the twin benefits of labwork preparing you for coursework (developmental genetics/PCR and DNA sequencing in my lab prepped me for several upper division cell/molec courses) and the coursework helps you understand the lab material in better depth.
Long story short: try your best to understand the WHY of the research. Even if it's not in your chosen field of interest, showing that you understand the research--and, equally important, that you can explain it to others--is an absolutely invaluable skill.
Fifth, your statement of purpose and personal statement. These are the essays you write to sell yourself to the university. All programs (at least that I'm aware of) require the statement of purpose (SoP), which normally consists of why you want to pursue graduate studies, how you've prepared for this, and why you feel that their program is a good fit for you (more on this in a moment). The personal statement is more, well, personal: generally, you'll talk about any difficulties you've had to overcome, personal reasons for pursuing a PhD, etc.
For example, a Statement of Purpose might read, "I'm pursuing graduate studies in order to gain a further background in my field, skills I'll need to pursue my own questions, etc," whereas a personal statement might read, "I was instilled with the desire to pursue neuroscience after my grandfather died of Alzheimer's." (Poor examples, but hopefully you get the picture)
Sixth, if you're looking for graduate schools, fit is normally the best indication. Right now, I wouldn't worry too much about it--you've still got a couple years to figure out your research interests and where they lie. When you're considering schools to apply to, though, the most important thing is fit. Prestige is nice (and by all means apply to prestigious schools), but make sure that the professors (3+) are doing research that you'd be interested in. It doesn't have to be a narrow, defined field; but just make sure you could see yourself happy in several labs if your first two advisors don't work out.
What I did when I was researching schools is used the US News Rankings (
http://grad-schools.usnews.rankingsandreviews.com/best-graduate-schools/top-science-schools), started at the top of the list, and worked my way down it. I would find a school in an area I might find appealing (I kept an open mind, but certain areas--such as hot and dry arid regions--would absolutely drive me nuts), go to the departmental web page, and then click through the faculty to see the research being done. If you're interested, when you get closer to applying, e-mail the professors and ask if they're currently accepting graduate students into their labs; responses may vary, but some might be interested.
Seventh, see if the programs have any open houses. I got SUPER lucky because the Michigan PIBS program does a PREVIEW weekend, and I was lucky enough to be invited to attend--they flew us out, let us meet and visit with faculty, showed us the school and area, and answered any question we might have. There's no better way to gauge a school than to simply visit it.
Finally, make sure you know what you're getting into. Do you want a master's, which would help in industry and enable you to help with somebody else's research? Do you want a Ph.D., which would enable you to perform independent research? Look at what each entails, and speak to a few professors before making up your mind (I was dead-set on a Master's when I first started; and now I'm (hopefully) going for a Ph.D. next year). Both have their pros and cons, and you can tailor them to suit your own needs.
If you need/want any other help, shoot me an IM on AIM or Yahoo: my screen name is Enresshou. I've been focused on this for the past six months or so, so I'd be glad to help out.
