The concept that changes in the dynamic behaviour of a rotor could be used for general fault detection and monitoring is well established. Current methods rely on the response of the machine to unbalance excitation during run-up, run-down or normal operation, and are mainly based on pattern recognition approaches. Of all machine faults, probably cracks in the rotor pose the greatest danger and research in crack detection has been ongoing for the past 30 years. For large unbalance forces the crack will remain permanently open and the rotor is then asymmetric, which can lead to stability problems. If the static deflection of the rotor due to gravity is large then the crack opens and closes due to the rotation of the shaft (a breathing crack), producing a parametrically excited dynamical system. Although monitoring the unbalance response of rotors is able to detect the presence of a crack, often the method is relatively insensitive, and the crack must be large before it can be robustly detected. Recently methods to enhance the quality of the information obtained from a machine have been attempted, by using additional excitation, for example from active magnetic bearings. This research is directed towards the concept of a smart rotating machine, where the machine is able to detect and diagnose faults and take action automatically, without any human intervention. This paper will consider progress to date in this area, with examples, and consider the prospects for future development.