I'm currently reading Steve Krug's new book - Rocket Surgery Made Easy. Peppered through it are his top usability testing tips - his "maxims".
A morning a month, that's all we ask. Steve's advice on the frequency of usability testing. Nothing big or clever, just enough to keep everyone focused and to keep usability on the agenda.
Start earlier than you think makes sense. It's never too early to consider some form of usability testing. Even the earliest concepts drawn on the back of a napkin. Personally I prefer the back of an envelope.
Recruit loosely and grade on a curve. Don't get too hung up on who you recruit to test with. Recruitment of participants can be a pain, after all. While it's better to have representative users, if you can't get them test anyway and grade your findings accordingly.
Make it a spectator sport. The more people involved with the website you get to watch, the greater the buy in you will get for usability improvements. And the reports and summaries you write after the test can be shorter.
Focus ruthlessly on a small number of the most important problems. The most important problems will bring the biggest benefits when fixed. Unless you're starting really early with your testing, there are going to be things you need to let go. Personally, I like to try and fix the easy ones too, even if they're more minor problems.
When fixing problems, always do the least you can do. This is the one I need to pay most attention to. I probably think too hard about fixes and strive for perfection. Steve says it's better to get something in place, and move on. There'll be more testing and more issues to uncover. Striving for perfection makes for a bigger job likely to take longer.