Let's assume for the sake of discussion that you are (or can enlist the services of) an excellent speaker and editor and you're able to cut out all unnecessary material, which is depressingly far from the truth with most tutorial videos and "screencasts" which tend to be full of stumbles and pauses and monotone blathering.
The answer is:
Two and a half minutes.
What's that number based on? I hear you asking. You just pulled it out of your backside, didn't you? Well, I'll tell you where I got it from.
Two and a half minutes is the average length of a YouTube movie.
You could not possibly ask for a better pool of data. We are talking about over 50 billion videos and over 2 billion views per day. And averaged out of all of these people and all of these videos, the majority are watched for well under 3 minutes, which is only a minute short of the average song length and about the same length as a typical commercial break on TV.
Some studies show that a typical attention span is 10 minutes, but that statistic is for college students in a classroom environment. That is, they are trying to focus their attention. Most users who watch your tutorial video are already in the mindset that this video is just a nuisance, it's something that they have to get through in order to start using some other thing that was too complicated and unintuitive to understand without a tutorial video. They desperately want to get through it as quickly as possible and get back to their "real" work or activities.
Two and a half minutes. That's it. More than three minutes and you're going to start losing people.
If you've got a lot of material to get through then split it up over multiple videos. One concept or task per video. Do not make people sit through a 10-minute video just to learn a single concept.
Attention spans seem to be declining every year with the advent of services like SMS messaging and now Twitter. Row against this current at your own peril.