How robust is C#'s pseudo-random number generator?

Anyone have a reference for a robust analysis of C#'s uniform pseudo-random number generator? I am writing something that uses the Random class and want to make sure it satisfies marsaglia / bruce schmeiser's criteria for RNGs. I don't even know if it uses mersenne twister, for example; I was hoping since C# is popular enough I might save some time using the legwork of others. If nothing is out there, I might put together an analysis myself.

Any help is appreciated. Thanks, Myles