I think its good to start with a bit of introspection. Once you are clear about how you work, then you can identify the sort of supervisor that will be 'best' for you. And you will be able to ask specific questions (of them or their current students) about how they work. e.g. do you like to get on by yourself and only show something when you've got a polished piece of work, or do you like to talk things over and get feedback as you go along? So would a good supervisor be one who waits for you to come to them and doesnt harrass you, or one who pops into the lab every day to ask how things are and insists on regular meetings every week?
Oh, and gut feeling is good too. in the interview, you are interviewing them as much as them interviewing you, and if it just doesnt feel right (or v.v.), then I'd go with that feeling.