Meijler et al. have recently challenged the classical concept of AV nodal conduction (the conduction hypothesis) and suggest that the AV node might be controlling ventricular rhythmicity through its automaticity electrotonically modulated by atrial excitation (the modulated pacemaker hypothesis). This article critically evaluates the three major arguments of Meijler: (1) the absence of convincing evidence for conduction of excitation in the AV node; (2) the prevalence of disproportionately short AV intervals in larger animals; and (3) elimination of RR intervals shorter than the cycle length of ventricular pacing during atrial fibrillation, to judge which of these two hypotheses would more satisfactorily explain various experimental and clinical findings accumulated in the past. Previous observations including microelectrode mapping of the rabbit AV junction during regular sinus rhythm as well as second-degree AV block, clinical and experimental studies on concealed conduction, and studies on the ventricular response to atrial fibrillation appear to be compatible with the conduction hypothesis, whereas clearcut evidence for automatic impulse formation in the AV node has not been presented, except in a small number of hearts showing spontaneous AV junctional rhythms. In view of these observations and theoretical considerations based on comparative anatomy of the AV node-His-Purkinje system and on the latest experimental study on the equine AV node, the authors conclude that the conduction hypothesis appears to better explain all the available data, except perhaps in a few cases with second-degree intra-AV nodal block.