For City of Heroes, we were lucky enough to have a community of fans who were very excited to give us feedback on the game, but having a pool of people you trust to give feedback is very important.
I'd usually roll things out to as few people as possible because getting a few hundred people reporting the same bug is a waste of time. Once you started to get around a thousand people betaing, the noise outweighed the value so it was almost a promo at that point (we never had the heart to boot people from the beta and forums).
Analyze feedback. This is important and time consuming. Make sure you have feedback gathering mechanisms in place. Forums work well, but instrumenting your code is important too. This can take a lot of time, and you usually think of the most important questions to ask after you wish you'd instrumented.
Another easy mistake to make is rolling buggy stuff out. It is a complete waste to show anything to people if you know it is broken. Even when you think it is solid, hallway playtest with a single person (or small group) and just watch them play. Right now that's all I need for Neverwinter, when I start to get lots of conflicting advice and I'm unsure of what to do next I'll start rolling it out to more people.
Finally, (and a little off-topic, sorry!) trust your gut and your vision. People are great at telling you something is wrong, but you almost never want to listen to their suggestions. You understand the big picture better than they do.