Rosenblueth's hypothesis states that atrioventricular (AV) nodal conduction delay and Wenckebach periodicity of AV transmission are not due to overall decremental conduction within the AV node but are due to a single step delay which is caused by a special element or layer of the AV nodal tissue. This paper discusses some theoretical considerations which allow detailed evaluation of the original hypothesis. Two artificial conduction structures which incorporate the Rosenblueth phenomenon are presented and tested by theoretical experiments that consider the potential of these structures to produce (a) basic pattern of Wenckebach periods, (b) decremental shortening of RR intervals during Wenckebach periods. These experiments are also employed to test whether or not the Rosenblueth concept can be used to explain (c) appropriate dependence of AV conduction changes on the prematurity of atrial depolarizations, and of (d) alternating cycle lengths such as may be seen with atrioventricular reentrant tachycardia. The results of the theoretical considerations show that the original concept of the Rosenblueth hypothesis is sufficient to explain (a) but it cannot be used for realization of (b), (c) and (d). A modification of the original concept complying with both (a) and (b) is proposed. This modified structure can also reproduce (c), but not simultaneously with (b). The experiments show that anisotropy of intra AV nodal conduction may create an electrophysiological mechanism of single-step delay. Different anisotropic conduction structures have to be considered to reproduce phenomenon (d).