The principal function of a protection relay is to protect the electric power system (EPS) from intolerable phenomena, such as fault produced by lightning, which can produce deleterious effects on EPS. Both impedance estimation protection and transient-based protection have become the most used protection relays for the lightning stroke classification. However, it is clear that currently these techniques suffer from drawbacks, which cannot correctly classify lightning stroke signals with speed and precision balanced, and cannot correctly distinguish the lightning stroke with and without a fault. Therefore, in order to avoid maloperation, protection schemes must be able to classifier lightning strokes correctly. This paper presents a novel methodology to provide effective lightning transmission-line protection. Principal component analysis is used to extract different patterns for lightning strokes due to shielding failure and backflashover. This work discloses that the transient voltages projected on the principal component coordinates clearly illustrate lightning strokes with and without a fault. In this manner, two pattern classifiers, based on spectral energy and artificial neural network, are useful for effectively classifying lightning stroke types. On the other hand, to evaluate the robustness of the proposed methodology, this research is also compared with the technique based on the wavelet transform, which has been the major tool used for the analysis of lightning strokes. Results show that the methodology proposed is fast and effective, which is not affected by operational conditions. Thus, this methodology can be applicable for high-speed protection systems. © 1986-2012 IEEE.