I think one of the things that really sets good presenters apart from poor ones is what they do when something goes wrong. A poor presenter:

• needs all their cycles to try to figure out what went wrong, and has none left for looking after their audience
• is focused on making the demo work and sticking to the original plan
• is rattled by the experience so that whether the demo works in the end or is abandoned, the rest of the talk is lower quality

A good presenter:

• has rehearsed the demos many times, so that most "boom!" moments have been seen before and can be fixed quickly
• doesn't need as much energy to look after the audience, so is more likely to be able to do it
• is focused on making the talk work
• has backup (screenshot of the result, an exe that was built earlier) so that something can be rescued
• can get through the failure quickly and get back to the flow so that the talk as a whole can go well

I linked a while ago to a picture of Steve Teixeira dealing with a blue screen. Now Brad Abrams has highlighted Bill Buxton (who I quoted a few posts ago) dealing (at Mix) with hardware that refused to co-operate. I aspire to do as good a job dealing with demo failure. Brad includes some other "demo failures at Mix" in his post, too. 

A tip that has served me well over the years: have a stock of optimistic "I am not an idiot" sentences to use while you're either giving up on the demo or doing what you need to do to make it work. "Hey, if it was perfect, we'd be shipping it" is good. So is "told you it would be a short demo". Humour keeps the audience with you, and stock lines don't take up much of your brain, so you can be furiously thinking with most of your brain about how to solve your problem (either how to fix the demo, or what to do with the rest of your talk now that you have ten more minutes to fill.)

Kate