How can I improve my chances of being accepted into graduate school for Theoretical Astrophysics?

Step one is the obvious: get good grades. More than that, make sure you really learn the subjects and how they interrelate; it can be easy to get good grades on homework and tests because the material is fresh in your mind, but you need to keep practicing and refreshing the knowledge to really make it stick.

Take some courses in related fields; computer science is an obvious candidate (learn a standard language like C/C++ or Java) since so much of science requires computers for control, simulation, and/or analysis. "Electronics for physicists" (e.g., Horowitz and Hill's Art of Electronics) is also excellent background even if you stick with theory. On the math side, you'll have to take calc, differential equations (ordinary and partial), etc., but also look into probability, statistics, complex analysis, group theory, and maybe even topology. For things like stellar evolution or planetary formation, some chemistry might even come in handy.

By the summer of your sophomore or junior year, try hard to get an internship somewhere where astronomy or physics is happening. You're almost guaranteed not to get something theoretical, but hands-on experience in a government lab or a university experiment or an observatory is invaluable, particularly since theorists almost always work hand-in-hand with observers or experimentalists. Knowing how they work and the limits of what the current and next generations of instruments can do will help guide your own research, and you'd be surprised how much hands-on, back-of-the-envelope physics takes place in such settings (and how useful it will be to learn to do BOTECalculations yourself). Do two or even three if you can manage it.

And don't be surprised if you find your interests have shifted by the time you actually get there. I happened to stick with theoretical (and computational) astrophysics through grad school, but I found to my surprise that near-field optics, various forms of atomic-force microscopy, and even sonic microscopy were quite fascinating in their own right. Adaptive optics for very large telescopes were (and remain) very topical and interesting, and the last few years have seen the birth of neutrino and gravity-wave astronomy, which are also very interesting and which make use of completely different technologies from photon astronomy. Even IR telescopy gets ridiculously cool when it's used to observe the orbits of the three dozen or so stars around the Milky Way's central black hole. (Two of them have completed full orbits in just the two decades that we've been looking, by the way.)