Re-entrant arrhythmias are the major cause of death from cardiovascular disease. A number of models or mechanisms have been proposed to explain the generation of re-entrant arrhythmias in the ischemic or damaged heart. However, none of these models can qualitatively predict the formation of re-entry movements with no modifications of the basic electrophysiologic characteristics of the myocardium. In this presentation we evaluate the concept that the generation of re-entrant arrhythmias is due to increased variance in the conduction characteristics of the cardiac tissue, rather than to modification of these properties. Using a model of a homogeneous two-dimensional matrix of excitable conducting cells, we derived the relationship between the relative standard deviation (RSD) and the probability of occurrence of a local ordered dispersion of velocities shown to have the potential to result in circular propagation. This probability was found to be insignificant when the RSD was lower than 10%, but increased dramatically for RSDs greater than 10%. On the basis of experimental RSD, the calculated probability for circus movement formation is one in 10,000 normal heart beats and one in two ischemic heart beats. The model provides new insight into the mechanism of re-entrant arrhythmias as well as new tools for diagnosis.