Single and dual chamber pacing algorithms have been incorporated into a realistic computer model of cardiac electrical activation. The model enables different pacemaker algorithms to be tested, it allows prediction of their behavior, and it produces a simulated ECG record for each case. The computer model has been used to test eight different modifications of a simple DDD mode to prevent or terminate pacemaker-mediated "endless loop" tachycardia: (1) constant prolongation of the atrial channel refractory period; (2) prolongation of the atrial refractory period after a ventricular premature beat (VPB); (3) atrial pacing synchronously with a VPB; (4) simple rate control; (5) rate control in which the VA counter is not reset; (6) no ventricular pacing after an atrial premature beat; (7) rate limitation of atrial sensing; and (8) a combination of DDD and high frequency atrial stimulation modes. These modifications were tested with VPBs, atrial premature beats, atrial stimulation without capture, and accelerating sinus tachycardia. Only the pacemaker designed not to pace the ventricles following an atrial premature beat behaves satisfactorily in all four circumstances. Further possibilities for the development and use of a pacemaker-oriented computer heart model are discussed.