Implantable cardioverter defibrillators reduce the risk of sudden cardiac death in patients with ventricular tachyarrhythmias. However, for the few patients with unacceptably high defibrillation thresholds at implantation the risk of sudden death may remain high. If a small number of defibrillation attempts are used to determine a defibrillation threshold, then a high defibrillation threshold may occur in some patients due to the probabilistic nature of defibrillation: a small percentage of shocks will fail even at optimal shock strengths. Basic investigations have suggested mechanisms for high defibrillation thresholds in other patients. The extracellular potential gradients produced by a shock correlate with ability to defibrillate and may be used to classify mechanisms for high defibrillation thresholds. Computerized mapping studies have demonstrated that extracellular potential gradient fields produced by defibrillation shocks are uneven with high gradient areas close to the electrodes and low gradient areas distant from the electrodes. A high defibrillation threshold may occur because: (1) a shock creates a subthreshold potential gradient in the low gradient areas; (2) a patient has a higher minimum potential gradient threshold than other patients; or (3) a shock leads to refibrillation in the high gradient areas. This article reviews experimental evidence to support each of these three possibilities then suggests experimental and clinical investigations that may clarify the causes of high defibrillation thresholds in patients.