This paper presents a novel method to identify inrush current and internal fault current using the two-instantaneous-value-product algorithm to extract the variation feature of the fundamental current amplitude. First, the two-instantaneous-value-product algorithm is developed, and different variation trends of the fundamental current amplitude for inrush current and internal fault current are analyzed. Then, according to the descending features of the fundamental current amplitude, the inrush current and internal fault current can be distinguished from each other. A total of 216 experimental measurements have been tested on an YNd11 connected transformer. Dynamic testing results indicate that this method is able to clear internal faults, even light ones, within a cycle and is not affected by the current transformer saturation. Moreover, compared with the second harmonic restraint principle and the waveform comparison principle, the proposed algorithm has better performance. The computational simplicity of the proposed scheme enables its implementation in real-time applications with low-cost microprocessors.