Abstarct: Due to the WHO (World Health Organization), the cardiac conditions and diseases dedicate majority of death causes among the diseases to themselves. Electrocardiogram has a wide usage area in the field of detecting heart diseases. Due to its cheapness, easiness and efficiency, its usage and knowing how it is working and also correct understanding and analysing of Electrocardiogram will help us in diagnosing the disease. Artificial methods can be used to reduce the cardiology specialist’s mistakes. Change or tilt in each of the electrocardiogram’s main signal could be a result of a disease. In this thesis a new algorithm baxsed on Walsh transform for classification of cardiac arrhythmia has been suggested. The main purpose of this research is to approach an automatic, efficient and fast way for diagnosing of cardiac disorders such as “RBBB”, “LBBB”, “pace rhythm”, “Premature Ventricular Contraction” and “natural mode” from each other. Walsh spectrum is used for feature extraction and feature reduction. Appropriate features from Walsh spectrum have been selected and for classification stage SVM method has been applied. Experimental results which have been done on a standard databaxse “MIT_BIH” show promising result.