The shear strength of beams and one-way slabs has been acknowledged for more than one century as one of the most complex, yet fundamental, topics to be addressed in structural concrete design. The experimental data used to investigate the phenomenon has traditionally been obtained from tests on simply supported beams subjected to point loading and recording only a limited amount of data (applied load and deflection in most cases). Following these experimental evidences, design formulas have been calibrate-din many cases in an empirical manner-by using large amounts of experiments in the form of databases. Also, mechanical models have been proposed as a rational approach to shear design, which can be conceptually very different and relies on different shear-transfer actions. Discussions on the pertinence of these models are mostly based on the general agreement of the failure load to existing datasets more than on a critical review of their basic hypotheses. Within this context, the development of new measuring techniques, such as digital image correlation (DIC), enables a systematic and transparent examination of the shear crack development and of the role of the various potential shear-transfer actions. This allows for a scientific discussion on the correctness of the principles of the design approaches and allows understanding their pertinence and limitations. Based on these possibilities, the main hypotheses of the Critical Shear Crack Theory (CSCT) for shear design are examined in this paper. To that aim, the results of a specific test series on specimens tested under realistic loading conditions are reviewed. The results show that shear strength results from a combination of various shear-transfer actions depending on the development of shear cracks and their associated kinematics. On this basis, the applicability and pertinence of the CSCT main hypotheses are discussed and a complete mechanical approach is formulated. Also, it is shown that the CSCT can be formulated in a simple and consistent manner as closed-form design equations.