A fault management system is operated for an access network which forms part of a communications network. In the access network, terminating lines in the form of pairs of copper wires extend from a local switch through a series of nodes to terminal equipment provided for users of the network. The fault management system includes a test head and an access network system. Each night, the test head performs a series of tests on each of the terminating lines. The results of the tests are transmitted to the access network management. The test results are then analyzed with respect to a set of parameters to identify characteristics that would indicate that a fault is likely to occur on the associated circuit within a predetermined period e.g. 1 year. Further analysis can then be carried out to establish the probability of the fault actually occurring and/or whether the potential fault analyzed is going to occur in either the underground or the over-ground part of the network.