I love making recipes, working usually within the basic confines of what the style is supposed to be like (malty, hoppy, dry, sweet, etc.). Sometimes, like yesterday, I just have a basic idea and work with it (a malty pale ale with lots of fresh hop flavor but not a lot of bitterness).
But - there's a lot of work involved. There's a lot of reading (Ray Daniels, Jamil's new book, lots of others). There's lot of reading about different malts and grains and hops, lots of trying different things and reading about people's experiences. It's not something you just *do* without putting in some effort.
Start with
www.howtobrew.com, like we suggested. Read it all the way through. Then read it again. Get a hard copy and put it in the bathroom (mine has about a half-dozen homebrew tomes in there). Try different beer styles (commercial examples) and read different reviews on BeerAdvocate or RateBeer, so that you can start to develop your pallate (maybe someone notes a "nutty" flavor that you didn't notice the first time). Look at clone recipes of beers you really like.
Basically, you build up a big database of information in your head about all the ingredients and all the styles and how all of that stuff can fit together; then, when you start putting your own recipes together, you've got a foundation to work with, you're not just throwing a bunch of random crap in the fermenter.
Here's a secret, too.... the recipe is often, I think, the least-important element of what makes a good beer. Think of a hefeweizen as a classic example - there's really only a couple of ways to put the recipe together, but they don't all taste the same. When you see guys like Mike McDole and Jamil Z consistently win competition after competition, it's not JUST that they have good recipes; it's that they control things like fermentation temperatures so well, and have such great processes in general, that their recipes can truly reach their full potential.