My tips (and I am far from a regular at this, having only run a couple pickup games at GenCon, but I'm feeling a bit helpful...take them for what they are worth):
1. Pregenerated characters - unless you really want to show off character generation, you should have pregenerated characters. Ideally, you'll use the characters to show off the variations in character concepts or whatnot.
2. The Most Fun Power Level - find the sweet spot in the system you're showing and make the adventure for that level (or the equivalent). The players will have more fun if you show them the game at its most fun and you'll be able to highlight the strong points of the game.
3. Chugga-Chugga Choo-Choo - con games should be on rails, as you will most likely be limited in the time you can spend and the last thing you want is a table full of players struggling with "what to do next". The players know that this isn't a campaign, so having a lot of freedom shouldn't be something that they expect. Also, you are trying to display a new system, not make the players direct the action. EDIT: This is too constricting. After reading the Piratecat's first post in the thread that he linked to and remembering my games, I'd like to revise this to match Piratecat's Narrow-Wide-Narrow idea. It's almost exactly what I did. I started them on rails, let them go far afield and brought them back in the end. Both of the games that I ran were the same adventure and both played entirely differently, with the exception that the beginnings were very similar and the ends were somewhat similar.
4. The Adventure Outline - my preference is to break my game into 3 or 4 manageable and easily separated parts (each one taking a roughly equal percentage of the total time). Each part will have its own little subplot and may show off a different aspect of the game. Maybe in Part I, the group finds itself shipwrecked on an apparently deserted island. Allow for a bit of "figure stuff out" time to see if the group will scavenge items or make a plan (I call this a time-filler and I will have removable parts to my game that I will use or discard, depending on how long the party takes with any "free time" they are given). Soon, they are attacked (show off the combat system). After the combat, let them "figure out the next step" (perhaps you have a clue lying around), or if they've used up a quarter or a third of the time (depending on the number of Parts in your game) show them the carrot. You can move throw each part of your adventure and easily fill a standard 4 hour time slot.
5. Shoot for the most important stuff - you will probably find that you aren't able to show everything, so make sure that you note the most important aspects of the game (be they class abilities, the combat system, skills, action point mechanics, aspects, traits, or whatever your game features). If you miss something, don't worry about it. Perhaps you can leave some wrap-up time at the end of the session to garner the players' opinions and relate anything important that you may have missed.
From what I've seen in playing and running games, the actual plot isn't as important as the GM's ability to keep the game moving and let the players do cool stuff. It's all about fun. If there is an aspect of the game that you think you may stumble on, leave it out. You can discuss it later, but you don't want to bog a game down.
Be loose. Keep the game moving. Have fun.
Well, I don't know how helpful any of this is, or how clear my post is, but I hope that you find it helpful.
