The two fatal ventricular arrhythmias, Ventricular Fibrillation (VF) and Ventricular Tachycardia (VT), are better treated using different electrical therapies: a lower energy cardioversion for VT and a defibrillation shock for VF. Automated External Defibrillators (AED), whose use for children is recommended since 2003, should discriminate between VT and VF in adult and pediatric patients. We propose a new method to discriminate VT from VF that applies sequential hypothesis testing to the wave mode sample entropy values of the ECG. Wave mode sample entropy values are calculated every second using overlapping 4 second signal windows. The algorithm was designed using a development database of 154 adult cases, 53VT/101VF, and then tested on two independent databases: an adult database of 92 cases, 64VT/28VF, and a pediatric database of 78 cases, 49VT/29VF We obtained an overall accuracy of 96.8%, 94.3% for VT and 98.0% for VF, for the development database. The accuracies obtained for the test databases were: 94.6% for the adult patients, 93.8% for VT and 96.4% for VF; and 94.9% for the pediatric patients, 93.9% for VT and 96.6% for VF. The algorithm accurately discriminates VT from VF in adult and pediatric patients.