In my experience, Fortran is widely used, and may, in fact, be what is most widely used in (astro)physics.
However, my belief is that it is inertia that is causing this. Fortran has been slowly dying for a while. The only fortran code I've worked with is something originally written in Fortran 77 about 25 years ago (I think) and updated by a guy who re-did it in Fortran 90/95 a few years ago. However, C and C++ is used more and more, particularly (I hear) in particle physics with data analysis and simulations done in root (i think that's the name). In my own work with x-ray timing, I almost exclusively use C, with a smattering of perl and other scripting stuff.
That said, if you can learn both Fortran and C very well, you should be set for a whole hell of a lot of things. You should also know unix well enough to do basic troubleshooting, since that is the environment (imho) most conducive to scientific computing, and I believe this opinion is backed up by the fact that most places who do a lot of it use unix-based systems.