A performance analysis of a networked control system (NCS) based on a controller area network (CAN) is analyzed. The network-induced time delay is stochastically modeled. The time-delay behavior is characterized by a hidden Markov model (HMM) framework. This model captures the variation of some CAN parameters such as network load, baud rate, and priority. A HMM-based estimator is included in a NCS to predict future process states. This proposal allows traditional control techniques to be applied in NCS with stochastic time delays. The approach was proved under different network configurations based on scale-model Full CAN X3 with real components of the Peugeot 807 vehicle.