This report covers the theoretical development of the safety state model for railroad operations. Using data from a train control technology experiment, experimental application of the model is demonstrated. A stochastic model of system behavior is developed which is used to estimate the dynamic risk probability in a human-machine system. This model is based on a discrete Markov process model. Based on observer behavior of an existing system, the model is used to determine an instantaneous risk probability function, which is dependent on the system state.