What are you majoring in? And what are your aspirations?
If you're in engineering/applied science, and you don't have grandiose plans of grad school, I second the suggestion of numerical methods. It'll give you some idea of how differential equations (which is what most physics equations are) are solved realistically.
If you're planning on grad school, I would highly suggest partial differential equations. Its the logical "next step"... and, as it turns out, a bit more practical than standard diff eqns. A STRONG grip on theory is essential before you commit to serious numerical solutions (which you'll address in grad school).
If you're in math, just take something fun... something that interests you.
Best of luck!