A distributed simulation system is provided in which timesteps may be divided into a first phase (referred to as the zero time phase herein) and a second phase (referred to as the real time phase herein). In the first phase, each distributed simulation node in the system may process one or more received commands without causing the simulator to evaluate the model in that distributed simulation node. In the second phase, each distributed simulation node may cause the simulator to evaluate the model in response to a command supplying one or more signal values to the model. In one embodiment, the second phase may iterate the evaluation of the model for each command received which supplies signal values. Each iteration may optionally include transmitting a command including the output signal values produced by the model during that iteration.