The detection and diagnosis of faults in sophisticated machinery systems presents the monitoring engineer with some of his most challenging problems. One of these is inter‐fault definition, first of all between the various components of the machine and then down at component level. For example, the intermediate casing of an aeroengine gas turbine is literally packed solid with shafts, bearings large and small, gears, splines, seals, etc., each with its own particular set of failure modes. In particular, the gears and bearings can fail in a large variety of ways, some more disastrously than others. For reasons connected with the statistical validity of his result, which is essentially to give the machine a “bill of health,” the engineer must seek ways of identifying the faulty state of each component as though it were an isolated unit running on a test stand. The author will describe what progress is being made in the U.K. to achieve this aim, emphasizing primarily the role of vibration analysis, but touching briefly on other important techniques such as debris, performance, temperature and electrostatic signature analysis.