Choosing a PhD (or postdoc) mentor
I just finished a really great article by Bruce Alberts (of Molecular Biology of the Cell fame and also Editor in Chief of Science) on how to choose a mentor. Dr. Alberts says "the exact PhD project is not nearly as important as finding the best place for learning how to push forward the frontier of knowledge as an independent investigator." i.e., learning how to be a good scientist is so much more important than the actual science performed. I have seen so many people who choose an advisor because they want to work on X, not recognizing that there will be many, many opportunities to work on X throughout their career, but really only 1 PhD mentor.
Dr. Alberts suggests that the best mentors take time to talk with their students teaching them not just how to do research, but also how to think critically about a project and present it to others from the proposal to research paper stage. When you are looking for a mentor, you should ask the students in the group how often they meet with the PI. Do they have career discussions with the PI? Have they had a chance to write abstracts, papers, grants? At what point in their training (at the very end or throughout)?
Choosing a mentor who can provide you this training is more important than almost anything else, really. You can change your research field by a few well-selected postdocs or by setting your own lab direction after your become an independent scientist, but you can never recover from poor or inadequate training. Just something to think about.