Since the existing fault phase identification methods can not identify all fault types quickly and accurately for EHV transmission lines, this paper proposes a new method of fault phase identification based on the fault component of phase voltage difference and the Kalman filter algorithm. The method defines the fault component’s ratio of one phase voltage to the difference of the other two phase voltages as a fault phase identification coefficient. By analyzing the characteristics of fault phase identification factors in each fault type, the fault phase can be identified. Simulation results show that using the Kalman filter algorithm to extract fundamental component is faster and more accurate. Meanwhile, the method can identify fault phases within half a cycle and is scarcely influenced by fault resistances, fault locations and fault initial phase angles. It also has a high sensitivity when the fault is on the side of strong source.