1. Address people by what they call themselves when they introduce themselves. Nothing says "student" to me like someone who calls me Dr. Academic after I introduce myself as Prodigal. My colleagues (and research group for that matter) call me Prodigal. My students call me Dr. Academic. If someone says "Hello, I am Dr. Pretentious", by all means call them Dr Pretentious. And definitely DON'T do the starstruck thing, even if you are meeting a personal hero or Nobel laureate.
2. When you talk about your research, own it. It is YOURS. Not your advisor's, not your program officer's, and not your university's. Also, don't refer to your advisor all the time. The interview is about you not her/him, and talking about him/her all the time will reinforce that you are a student/postdoc.
3. When you do talk about your advisor (and it will come up), don't act deferential. Be respectful, but it seems strange to me when interview candidates call their advisors "Dr. Advisor". Don't call your advisor "my boss" either.
4. When you meet with your potential colleagues one on one, and the topic turns to research, they want to have a discussion with you. Even if the person you are meeting with is a giant in your field, they do not want you to just accept their suggestions as the "word from on high". When a candidate does that instead of engaging in the conversation, it makes me feel like I am having a one on one with one of my own trainees, not discussing science with a colleague. Bonus points if you can offer something constructive about their own research!
5. When meeting with students and/or postdocs, remember that you are interviewing to be a faculty member. No matter how tempting it is to talk about stuff you got away with in grad school, or how annoying your advisor is, don't.
6. While "I don't know" is a perfectly valid response to a question (and certainly better than trying to fake it), another response is to try to think things through, or offer your thoughts about the topic (obviously, this isn't true of questions about factual things!). People who can pull this off definitely seem more scientifically mature.
I'm sure there are other things, but this is what comes to mind from my (admittedly) brief experience on the other side of the interview.