The Support Vector Machine (SVM) has been adopted here to identify four different Partial Discharge (PD) sources that can affect the insulation system of AC rotating machines. A number of Roebel bars were prepared to generate bar-to-finger, corona and slot PD in addition to the distributed micro-voids that are typical of this insulation type. PD measurements were performed using different set-up conditions, defect locations and voltage levels in order to produce examples of PD activity that represent the same source under a range of conditions. The SVM was trained to differentiate between the inherent features (global and derived parameters) of the phase resolved PD (PRPD) distributions produced by each discharge source. In order to achieve the optimum source classification accuracy, different combinations of distribution features were used to produce a range of SVM models to identify which parameters were influenced by the measurement conditions. A cross validation technique has been used to obtain the highest testing accuracy. Moreover, results obtained using raw data and normalized parameters, were also compared to obtain the best identification performance of the given defect typologies.