I think it indeed depends a bit on the academic system in your country. In Germany we have for some years the socalled "Bologna System" now (I don't want to go into the comparison to the old German system in this thread; it's too sad). Here the usual way is: You start studying Physics towards a Bachelor of Science (BSc) degree, which takes 3 years (organized in 6 semesters). Usually this includes both experimental and theoretical physics (mechanics including relativity, classical electromagnetics, non-relativistic quantum mechanics, statistical physics, and some physics lab work) as well as mathematics. Depending on the university the physics students either have to attend the mathematicians' linear algebra and analysis lectures or there is a special lecture for physicists. I'm a bit undecided which way is better. I studied physics in a university where one had to attend the mathematicians' lectures. I liked it, although it's not perfectly geared for what a typical physicist really needs, particularly in the beginning. The problem in the early semesters is that you don't have enough math to really do the physics right, but on the other hand you learn a lot of math also in an intuitive way in the physics lectures. The problem with a mathematicians' math course for physicists is that you don't get as much of the practical computational skills as you need for physics, i.e., you should be able to calculate an integral, while in the math lecture you rather learn to prove that the integral exists, why you can or can not commute a limit with integration, and so on. The problem with a lecture "math for physicists" is of course the opposite: Often you learn how to calculate things without really understanding what's behind such operations. So the best is to find some good compromise between mathematical rigor and the computational skills necessary to be a good physicist. It's also not true that a priori a theoretical physicist needs more math than an experimental one (btw. as a physicst you should not specialize too early in my opinion, because there's a lot of common wisdom any physicists should acquire, and then you can much better judge, what specialization is really interesting for you personally). It may only be true that an experimental physicist needs different math than the theoretical physicist in his or her later specialization (e.g., an experimental physicist may need much more statistics and mathematical error analysis of experimental data, while a theorist working, e.g., in General Relativity needs more details about differential geometry). In this sense, the most important thing you learn in the math lectures is how to read a math book to get the math you really need. At the end of the BSc curriculum you write a Bachelor's Thesis. This takes usually 3 month and in the best case may consist of some research project (in our theory group we even managed to get some BSc students to publish his or her first scientific paper in a peer reviewed journal!).
After you acquired your BSc you can go on (in my opinion you also should go on!) studying towards a Master of Science for two more years (4 semesters), where you have more lectures on advanced topics and, most importantly, to some research on your own under supervision of a professor (leading to a Master's Thesis).
Of course, after this, you can go further towards a PhD.
