I'm working in the field of experimental psychology (uh, better call it 'behavioral sciences', there is too much garbage in psychology...) at the MPI for Human Development (Germany). After writing a lot of code in C++ during the last years (programs for psychological experiments, statistical analysis, simulations, ...), I switched to C# last year. Development is much faster now, mostly due to the dramatically reduced time to find bugs or conceptual errors. Moreover, there is a bunch of technologies that I actually use now in my daily work (e.g. Managed DirectX, component development, XML/XSLT, GDI+, remoting, distributed computing, targeting mobile devices) but seemed - at least to me - sometimes too complicated to use with Visual C++ and MFC (yes, I'm not a computer scientist by education; and yes, I do think that COM is pretty complicated!).

To sum up: To me, scientific development with C#/.NET is not just simpler and faster than before. It gives also access to a huge toolbox that allows me to develop quite different kinds of more or less sophisticated applications that even look better than before (finally!).