I've been thinking lately about PhDs, so I've written down some advice for anyone considering starting one, based on my own experience and those of the students and former students that I know. My PhD was in neuroscience but as far as I can tell, the situation is similar in most sciences. However, I'm not sure how far the following applies outside of the UK. It's important to find a course and supervisor that's right for you - because your supervisor is pretty much your God. There are few checks and balances on their influence. If you don't get on with God, you're in hell. I know plenty of horror stories of PhD students who were mistreated by their supervisor. I was lucky enough to have an awesome one, but it really was 'lucky', because I didn't get to know her before I started. It could have gone either way. Don't fall into the trap of assuming that someone who's published lots of exciting papers is going to be good to work with. Scientific talent and basic human decency are, unfortunately, uncorrelated. This is why I'd recommend working with prospective supervisors before you commit to a PhD with them - either as a master's student, or as a research assistant. A few months will be enough to know if that lab is a good place to be. This is also a great way of getting onto a PhD course, because it teaches you practical skills (see later). You will know more science than your supervisor. In fact, you know more science now than you ever will again. During your PhD you'll forget most of what you learned previously (although if you do teaching, this will slow the decline some). The reason is that (except maybe in physics) the day-to-day business of research just doesn't involve the kind of knowledge you were taught. What this means is that you will not be expected to know lots of facts except in the narrow domain of the speciality you'll be working in. And even there, you rarely need to memorize anything, you just pick it up.PhD students are there to do, not to know. So when applying for a PhD place, you'll really benefit from knowing how to do things. If you're skilled at using an important method, or piece of equipment, or software, that will put you in demand. Some kind of coding (MATLAB, C++, R) is a big plus in many fields; in neuroimaging, knowledge of one or several of the big analysis toolboxes (SPM, FSL, FreeSurfer, etc) is also a big help. Getting good exam results is also important, but it's by no means everything. Hopefully if you've read this far, you'll have worked out that a PhD is very different from other degree courses. In a nutshell, while the main challenges for undergraduates are intellectual, on a PhD the challenges are emotional. Instead of wrestling with exams, grades and essays, you'll face self-doubt, frustration and boredom. I'm not trying to be negative: on a good PhD course, there will be plenty of awesome things as well. But inevitably there will be challenges, and they will not be the kind of thing you can overcome with intellect alone.