The term defibrillation threshold is usually understood to mean the shock intensity just enough to defibrillate a specified cardiac chamber (atria or ventricles). With the advent of so many different types of defibrillator, it is important to be able to specify the defibrillation threshold, which has frequently been described by the classical strength-duration curve. Another method of representing defibrillation plots the percent-successful defibrillation against shock-strength area. The mechanism of defibrillation is discussed, and the concepts of the strength-duration curve and percent-success against shock-strength curves are compared. Because defibrillation is associated with a time-varying spectrum of cellular excitability, a given shock strength will not always achieve defibrillation, and this produces the sigmoid shape for the curve that relates percent-successful defibrillation to shock strength. Therefore it is important to recognise two concepts: first, there is a family of strength-duration curves for defibrillation, each curve representing a given percent-successful defibrillation, and, secondly, there is a family of percent-success against shock-strength curves, one for each pulse duration. Canine ventricular defibrillation data are used to bring these two concepts together. Most importantly, the concepts adduced in the paper apply to transventricular, intracardiac and transchest defibrillation; the only difference in these applications is a scale factor that represents electrode location with respect to the heart.