Protection relays are designed and applied in power systems to protect equipment from excessive damage and to maintain system integrity and stability. The main operations of relays such as early fault identification, rapid equipment isolation and fault location are imperative to maintain the reliability and integrity of the system. However, traditional protection schemes, such as Transient Based Protection (TBP) and protection schemes based on impedance measurements suffers mal-operation due to the presence of lightning stroke on transmission lines. Therefore, in order to avoid mal-operation and delayed tripping, protection schemes must be able to discriminate lightning strokes correctly. With this purpose, almost all lightning transmission lines protective relaying algorithms use integral transforms such as the Fourier transform and the wavelet transform. However those techniques can only provide an average attribute of the signals or their components. This work, introduces Principal Component Analysis (PCA) to extract patterns of lightning strokes on transmission lines. Results show that lightning stroke can exhibit well-defined patterns which can be used to classification and location of lightning strokes. © 2003-2012 IEEE.