New implantable devices are now available that can offer different therapies for different arrhythmias but they need a method of discriminating between these rhythms. Heart rate analysis is predominantly used to discern between sinus rhythm (SR) and pathological tachycardias but this may be of limited value when the rates of the rhythms are similar. An enhanced form of Gradient Pattern Detection (GPD) has been developed using an 8-bit microcomputer that can distinguish between SR and up to three other arrhythmias in real time. This is a method based on electrogram morphology where each rhythm's specific electrogram is classified by a sequence of gradient 'zones'. The microprocessor of the computer is of similar processing power to ones used in current pacemakers. Five patients with multiple arrhythmias were studied. Four had ventricular tachycardia (VT) and one had three conduction patterns during supraventricular tachycardia (SVT). Bipolar endocardial right ventricular electrograms were recorded during SR and tachycardia in all patients. The computer would first 'learn' about each different rhythm by a semi-automatic means. Once all the rhythms were learned the program would enter the GPD analysis phase. The computer would output a series of real-time rhythm specific marker codes onto a chart recorder as it recognized each rhythm. Sixteen different arrhythmias (13 VT, 3 SVT) were examined for this study. All rhythms (including SR) were distinguished from each other except in the case of one patient with six VTs where two VTs had identical shapes and therefore could not be detected apart. The method would be a useful addition to heart rate analysis for future generations of microprocessor assisted pacemakers.