An efficient policing mechanism for Markov model-based traffic descriptors of VBR MPEG video traffic is proposed. A VBR video sequence is described by a set of traffic descriptors using a scene-based Markov model for effective resource allocation and accurate QoS prediction. To monitor the input traffic from the source, the authors have defined a set of a limited number of representative monitoring parameters and also defined an exponential bounding function in order to accommodate dynamic behaviour. The simulation results show that the proposed policing mechanism for Markovian traffic descriptors monitors sophisticated traffic, such as MPEG video, effectively and also protects network resources from both malicious and abnormally behaved traffic.