Understanding the rocking motion phenomenon is crucial to describing the dynamic response of masonry structural elements and assessing their actual vulnerability to seismic events. Although the first pioneering studies regarding rigid body rocking date back to the late 19th century, this topic still presents some points to be clarified. The analytical basis was laid by Housner [1], who proposed a simple but effective model for slender blocks subjected to various types of dynamic loads; later works demonstrate, using experimental results, that energy loss is overestimated by the analytical model and that the impact event needs to be studied more deeply. The nonlinear equations of motion for non-slender blocks have been later studied numerically but, due to the chaotic features of the motion, this topic is far from being fully understood. As it is well known [2], even small differences in the initial conditions may easily lead to very large differences in the rigid block motion, hence greatly complicating the analytical investigation. The present contribution is part of a broader research activity aimed at investigating the still open issues concerning the assessment of the dynamic response of masonry structures subject to earthquakes. Starting from a review, both analytical and numerical, of the motion of a single block, it is shown that, by a critical re-examination of the limit condition for the stability of the rocking body (i.e., the overturning condition, also called “zero strength configuration”), a wider admissible domain can be found for the initial conditions, which shape depends on the block size, slenderness and loss of kinetic energy during impact. A numerical survey of the sensitivity of the dynamic response to the main parameters characterizing the seismic input is performed, using a numerical code developed specifically for this purpose.