This paper presents method to discriminate between transient voltage stability and voltage sag. The discrete wavelet transform (WT) is a powerful tool in the analysis of the transient phenomena in power systems because of its ability to extract information in both the time and frequency domain. This paper introduces a technique for accurate discrimination by combining WTs with neural networks (NNs). The WT is first applied to decompose the signals into a series of detailed wavelet components. The wavelet components are calculated and then employed to train a NN. The simulated results presented clearly show that the proposed technique can accurately discriminate between transient voltage stability and voltage sag in power system protection.