Call admission control is a principal component for QoS delivery in IP networks. It determines the extent to which network resources are utilized. It determines also whether QoS are actually delivered or not. Continuing on the steps of the existing approaches, we introduce a distributed and scalable admission control scheme to provide end-to-end statistical QoS guarantees in Differentiated Services( DiffServ )networks. This scheme is based on the passive monitoring of aggregate traffic at the core routers, and active probing of network state between the different edge routers. Thus, we have simulated a DiffServ network where different users are provided different service classes. In this network, the back ground (BG or BE) flows are characterized by Poisson process, however the Expedited Forwarding (EF) flows are generated by Exponential and Pareto ON-OFF sources. Afterward, we have evaluated the admission mechanism employed at the different network ingresses by measuring the network utilization at the network bottlenecks, and the effect of this mechanism on the accepted number of EF flows and their delay budget.