The advances in Cyber–Physical Systems (CPS), vehicular networks and Intelligent Transportation System (ITS) boost a growing interest in the design, development and deployment of Vehicular Cyber–Physical Systems (VCPS) for some emerging applications. As one of the key application for realizing traffic guidance, the traffic prediction could provide better route planning for people and accuracy decision basis for traffic managements. In practice, short-term traffic information has the characteristics of real-time, incompleteness, non-linearity and non-stationary, and few proposed methods could successfully implement this forecasting. In this paper, we proposed a fuzzy Markov prediction model which can estimate the short-term traffic conditions in VCPS in urban environment. First, we selected a real-time GPS dataset in the Shanghai Transport Grid Project as our data source for traffic prediction and pre-process this raw dataset to make it consistent with the practical case. Next, we combine the fuzzy theory with Markov progress in the prediction model, and use the continuous three-step average method to reduce the errors caused by the one-step transition. Finally, we choose the speed and traffic flow to express the metrics of traffic state and use the fuzzy reasoning rules to give out the determined traffic state. The simulation results show that our proposed model can be precisely used for the short-term traffic prediction in urban environment. © 2017 Elsevier B.V.