Frequency response analysis (FRA) is a frequency-domain method which is used to detect mechanical faults in transformers. The frequency response of a transformer is determined by its geometry and material properties, and it can be considered as the transformer’s fingerprint. If there are any mechanical changes in the transformer, for example if the windings are moved or distorted, its fingerprint will also be changed so, theoretically, mechanical changes in the transformer can be detected with FRA. A problem with FRA is the fact that there is no general agreement about how to interpret the measurement results for detection of winding damages. For instance, the interpretation of measurement results has still not been standardized.The overall goal of this thesis is to try to enhance the understanding of the information contained in FRA measurements. This has been done in two ways: (1) by examining the FRA method for (much) higher frequencies than what is usual, and (2) by developing a new method in which FRA is combined with the ideas of Time Domain Reflectometry (TDR). As tools for carrying out the above mentioned steps, models for the magnetic core and the winding have been developed and verified by comparison to measurements.The usual upper frequency limit for FRA is around 2 MHz, which in this thesis has been extended by an order of magnitude in order to detect and interpret new phenomena that emerge at high frequencies and to investigate the potential of this high-frequency region for detection of winding deteriorations.Further, in the above-mentioned new method developed in this thesis, FRA and TDR are combined as a step towards an easier and more intuitive detection and localization of faults in transformer windings, where frequency response measurements are visualized in the time domain in order to facilitate their interpretation. / QC 20111122