This paper presents a new methodology based on the analysis and monitoring of the discrete wavelet transform (DWT) detail coefficients (DetC) for internal fault conditions identification and discrimination in power transformers. The method extracts the characteristics and calculates the DetC signals levels variation through DWT decomposition for identification of transients. During internal faults the DetC present a sudden and high variation, different from what happens in other transients, allowing their discrimination. The proposed method was able to correctly identify the different types of internal faults in power transformers and additionally transient as inrush currents. The main novelty of the method are: (1) Intelligent data analysis and monitoring, without a threshold adjustment; (2) Fast identification of an internal fault conditions and high coverage of turn-to-turn faults; (3) Secutity during transients such as inrush currents, external faults and overexcitation; (4) Simplicity. The results show that the proposed method has potential for real application in power transformers protection.