A method of online fault identification in rotor/magnetic bearing systems is presented using wavelet analysis. A filter bank approach is taken to identify the discrete time wavelet coefficients of the rotor displacement signals. From artifacts present in the discrete time wavelet series associated with specific faults, it is shown that it is possible to identify both the onset time and the fault type. This method is demonstrated for simulations of a flexible rotor/active magnetic bearing assembly during auxiliary bearing contact and direct synchronous forcing for a range covering flexible critical speeds. Experimental validation was performed on a flexible rotor/active magnetic bearing facility undergoing sudden rotor unbalance, resulting in rotor orbits with and without auxiliary bearing contact. Artifacts associated with both the sudden mass loss and the rotor/bearing contact are identified.