This is the first tip on this list because it's without a doubt the most important. When approached with a forty-eight-hour deadline to make a game, most people's initial reaction is to want to order a couple of pizzas, stack up on energy drinks, and lock themselves into a room for the entire duration. This is a bad idea.
As for exercise, it isn't absolutely necessary; I'm sure we've all spent two days at some point in our life without exercising, game jam or not. No matter how well-rested and fed you are, doing some light exercises and stretching your legs (indoors!) can really help out.
This tip is a simple one: if you're debating whether or not to make a 2D game or a 3D game, make a 2D game. Working in three dimensions adds an extra level of complexity to art and coding, and the payoff is almost never worth it. You may stand out more with a fully 3D game, but it probably won't play as well, and you'll probably have to spend more time working on basic functionality and much less experimenting with fun new mechanics.
When you're thinking up the design of your game, remember that simple mechanics that work well are infinitely better than complex ones that aren’t. Don't plan for multiplayer, complex A.I, cut-scenes or a physics engine. In fact, when designing your game, I recommend trying to come up with something you think you could easily complete in only twenty-four hours, rather than the full forty-eight. Things almost always take longer than you anticipate, and there's a good chance you'll only barely finish by the end of the forty-eight hours. If for some reason you get things done faster than you now have that much more time to polish your game, and polish is what separates the good from the great.
The more you learn to code the more you come to appreciate the importance of clear, well-structured code. You learn the dangers of massive nested blocks and the beautiful simplicity that comes with well-executed inheritance structures. Forget it. You only have forty-eight hours, and you need to get things done. Don't waste time setting up versatile classes with minimal redundancy, code for functionality instead. It might be difficult to understand your code when you look at it a week later, but you'll get the best results in the moment, and in a game jam that's what counts. If you want to continue your project afterward, you can fix things retroactively. It's a bit of work, but it's work you shouldn't be spending your forty-eight hours doing.
This might be common sense to the more experienced of you out there, but this is a huge pitfall that snares many a game jam newbie. With only forty-eight hours to do something, people are inclined to get to work immediately. This often means jumping on the computer as soon as the theme is announced and starting to pound out some code. Why do this? You don't know what it is you're making within a split-second of the theme getting announced and you certainly aren't going to figure it out as you go. Take some time to decide on exactly what it is you're going to try to make, have something specific in mind when you first hit the computer, and see what happens from there. For a more in-depth look at this topic take a gander at What to Do Before Even Touching a Computer (though keep in mind its suggestions might go a little overboard in the context of a game jam).
You're not the only one participating in any given game jam; there are going to be a lot of games, and the people playing them will have a lot to choose from. This is where distribution is important: with so many choices of games with such a wide variety of quality, players won't want to invest a lot of time into setting any specific game up. Ideally, your game should be playable one click away from your submission page. Obviously browser-based games are the best for this, but a standalone executable can work well too. Avoid requiring your player to download some sort of framework or plugin they aren't likely to already have, and under no circumstance have your game require an installation. If they see your game takes too long to set up, most people will simply skip over it and have a try at another game that gives them less hassle. Choose your development platform accordingly.
Period.