My current project is finally in playtest and I'm really pleased with how it's going.  I did the second session earlier today with a group of my son's friends, and it seems to be meeting my goals.  More important, the kids are having a lot of fun with it, which I think is a very good sign.  It's been really useful to spot things that are ill-defined and things that need to be added to the rules – usually because one of the kids has a question and I don't have an answer for it.  My question, especially for those of you who have run a playtest, is what did you do to get the most benefit from it?  I'm also interested in hearing about things that didn't work well, so I know to avoid them.

Don't try to get everything you can out of a playtest.  Try to focus on a few things.  There's a difference between an early, informal playtest versus ones that come later.  (Eventually, you'd like to hand it off to another group and let them report back, but until it works under less stringent conditions, no point in that.)

Things that have worked for me with informal tests:

- Targeted cheat sheets that focus on what I want to test.
- Whenever possible, instead of answering how something works, point to the location in the rules or cheat sheet(s).  (I realized this after a couple of tests that didn't go as well.)
- Keep the adventure simple so that you can focus on the test, though it does help to target the adventure on specific aspects that you want to test. 
- If you have multiple groups available (or some players joining a group later), make sure to take full advantage of this fact by cleaning up obvious problems that the earlier tests found.

Things that didn't work as well on informal tests:

- Pre-gen characters can test the game, but not the character creation.
- Character creation always takes longer than you think it will, because there is always something confusing or unorganized or simply novel that will throw people for a loop.  You've got to test that too, and the only way to do it the first few times is suffer through it. 
- Nevertheless, it's better to start with pre-gens, because people will find it easier to make characters after they know how the game works.
- Nevertheless, you've eventually got to find a few people willing to to do it the hard way.

Finally, some things I've found helpful on the process itself:

- Don't try to fix anything in the rules while testing.  Just make a ruling to keep the game going and a note that you did so.
- Rulings you make this way are sometimes better than your initial draft of the rules.  Listen to your instincts on this.
- Judge everything on whether it works as designed or not. Not to say that design goals can't change or be refined, but sometimes players don't like a mechanic because it sucks and sometimes they don't like it because that's not the way they'd design it.  You've got to separate those two pieces of feedback.

If you wrote it, don't be the one to run it for playtesting. Let someone else run it while you observe and take notes. Also, don't answer questions from the GM during the game.