Described herein is a method of controlling call admission for packet switched networks, each network including at least two local area networks (50, 60) and a connecting network (70). The method comprises transmitting a burst 5 of trial data of the same size as the packet to be transmitted from a first node (52, 54) in a first local area network (50) to a second node (62, 64) in a second local area network (60) via the connecting network (70). The connecting network (70) comprises a plurality of routing nodes (72, 74, 76, 78, 80) for routing the burst of trial data of the same size as the packet to be transmitted from a first node (52, 54) in a first local area network (50) to second node (62, 64) in a second local area network (60) via the connecting network (70). The connecting network (70) comprises a plurality of routing nodes (72, 74, 76, 78, 80) for routing the burst of trial data to the second node in the second local area network along a particular path. The burst of trial data is reflected by the second node in the second local area network (60) back through the connecting network (70) to the first node in the first local area network (50). A comparison is carried out between the transmitted burst of trial data and the reflected burst of trial data to determine an estimate of packet loss rate of the path, and if the packet loss rate is acceptable a transmission of a continuous stream of data is initiated.