The electrocardiogram (ECG) has broad applications in clinical diagnosis and prognosis of cardiovascular diseases. The accurate description for the question how the ECG come from the cardiac electrical activity is helpful for understanding the corresponding relation between the ECG waveform and cardiovascular disease. Experience is the primary method of studying the ECG, but the computer simulation method makes it more convenient to explore the effect of given factor for ECG waveform. Cellular automaton is a simple and effective computer simulation method. However, the cellular automaton model considering the main structure of the heart is not yet established. Therefore, we propose a cellular automaton model for the ECG considering the atria, the ventricle, and the ventricular septum. With this model, the conduction of the myocardial electrical activation is simulated by following the field potentials under healthy and diseased conditions, and the underlying mechanisms are analyzed. Through the computer simulations and analyses the results are obtained as follows. First, the conduction process of the electrical signal in this model is the same as that in the real heart. Second, under the healthy conditions, the behavior of the field potential appears as normal ECG, in which the P wave and the QRS wave group come from the depolarization of the atria and ventricle, respectively, on the other hand, the T wave and J wave come from the repolarization of the ventricle. The computer results support the conclusion that the J wave appears just because the existence of the notch in the epicardial transmembrane potential curve. Third, the endocardium ischemia conditions result in the T wave inversion. The mechanism is that the action potential duration of the ischemic endocardial cells is shorter than that under normal conditions, which makes larger the transmembrane potential gradient between the endocardium and the subepicardium, and then contributes a more negative value to the field potential. Fourth, the epicardium ischemia leads to the higher T wave, and this is because the shorter action potential duration of the ischemic epicardial cells brings in a larger transmembrane potential gradient between the epicardium and subepicardium, which makes the field voltage larger. Fifth, the T wave appears earlier under the through-wall ischemia. The action potential durations of cells of the endocardium, the epicardium, and the subepicardium all become shorter under the through-wall ischemia, then the repolarization processes of all of these three walls are ended earlier, which leads to the earlier T wave. The cellular automaton model proposed in this paper provides a reference for the further study of ECG.