Abstract: Accurate analysis of ECG signal is of utmost importance as the amplitudes and intervals value of ECG provide information about proper functioning of heart of every human.Therefore there have been numerous researches going on for analysis of ECG signal.In this paper we discuss about various methods that are used to extract features of ECG signal and further classify them into various disorders based on the features extracted.We also discuss about various preprocessing methods to remove base line wander and other contaminants from the ECG signal.Most researchers take input signal from MIT-BIH Data base.Performance of preprocessing is measured using Signal to Noise Ratio (SNR) and performance of feature extraction methods and classification is measured using Sensitivity and Positive Predictivity.