I think the most difficult thing to learn when starting out teaching is to know when it's okay to just say, "You know, that's a really good question, and I don't know the answer. If you'd like, I can look into it further and let you know next class," and when you're going to look like a total idiot if you don't know the answer. But, yes, as Shmoe already stated, students can sense fear, so if you really are stumped, there comes a point where it's better to just admit it than to keep trying to bluff. It does nobody any good for you to keep going in circles with a problem, and they'll lose all respect if you waste their time that way. I had a TA like that, would just keep going and going, and we could see he was just looping back to earlier steps and going absolutely nowhere. For the most part, we stopped attending class and just showed up on the days when there were quizzes. There was really no point for us to be there when he had no idea what he was doing and wouldn't even admit he was stumped.
The other thing you'll want to learn to do is to explain the same point in several different ways. If you only know one way to explain it, when that doesn't make sense to a student, you need to be familiar enough with the idea to rephrase your explanation so they will get it. I'd suggest presenting the material for your first class or two to someone else, and let them ask questions, etc., and have them give you feedback. When I started out as a TA, we did this. All of the TAs for the course met once a week too, and actually did the labs we were teaching so we knew where there were likely to be glitches. If that's not an option, you could always ask someone else to just sit in on your class and give you feedback on what you could improve, because there's always something to improve. Even seasoned lecturers can use a little feedback from time to time to remind them of things they've begun to forget to pay attention to when giving a class.