This paper presents a new approach of fault detection for EHV transmission lines using one-end data for either of voltage or current signals. The scheme is based on the use of modified wavelet transform and implementation of digital signal processing concepts for the experimental results. A modified wavelet function is used to analyze the high frequency components initiated by the abnormal conditions. The algorithm is evaluated using the alternative transient program ATP model of 500 KV, 300 Km transmission line. The obtained results indicate an accepted degree of accuracy for the suggested technique. In addition, the new form of wavelet function is implemented in the real time model to detect any sudden changes in the voltage signal due to fault occurrence through the line.