Electrocardiogram (ECG) signal processing is one of the earliest challenges for scientists as a tool for cardiovascular assessment. T-wave changes are one of the most common abnormalities noted on an Electrocardiogram (ECG). The analysis of T waves in the ECG is an essential clinical tool for diagnosis, monitoring, and follow-up of patients with heart dysfunction. This paper presents a discrete wavelet transform based system for the extraction and analysis of T wave. The features like amplitude, frequency, energy are extracted from T wave to classify them into normal or arrhythmic. Classification can be done by using artificial neural networks. The above wavelet technique provides less computational time and better accuracy for classification, analysis and characterization of normal and abnormal patterns of ECG.