We present a new fuzzing technique for multithreaded C# programs running on the .NET platform. It is built upon the .NET Profiling library, supported by CLR (Common Language Runtime) on Windows, and uses configurable strategies for the fuzzing process. During execution of the subject program, the fuzzing algorithm controls thread scheduling and preemption through suspending and resuming threads at specific code locations that we call stop points. For the purpose of driving the fuzzing process, we have designed a hybrid systematic-random strategy that gradually finds yet unexplored thread schedules. Results of experiments with programs from the SCT benchmark collection show that our tool is able to find errors triggered by specific thread interleavings, and within practical time limits.