It has been proposed that VF waves emanate from stable periodic sources (often called "mother rotors"). Our objective was to determine if stable rotors are consistently present on the epicardial surface of hearts comparable in size to human hearts. Using new optical mapping technology, we imaged VF from nearly the entire ventricular surface of 6 isolated swine hearts. Using newly developed pattern analysis algorithms, we identified and tracked VF wavefronts and phase singularities (PS). We introduce the notion of a compound rotor in which the rotor's central PS can change and describe an algorithm for automatically identifying such patterns. This prevents rotor lifetimes from being inappropriately abbreviated by wavefront fragmentation and collision events near the PS. We found that stable epicardial rotors were not consistently present during VF: only 1 of 17 VF episodes contained a compound rotor that lasted for the entire mapped interval of 4s. However, shorter-lived rotors were common; 12.2+/-3.3 compound rotors with lifetime > 200 ms were visible on the epicardium at any given instant. We conclude that epicardial mother rotors do not drive VF in this experimental model; if mother rotors do exist, they are intramural or septal. This paucity of persistent rotors suggests that individual rotors will eventually terminate by themselves and therefore the continual formation of new rotors is critical for VF maintenance.