This paper presents a critical comparison of the existing code provisions for the shear strength of concrete beams. The comparison is based on the computerized filtering-out of the inevitable statistical bias from the available multivariate database on shear strength, on an examination of the predicted size effects on shear strength and their underlying hypotheses and on the results of recent high-fidelity numerical simulations of shear failure. In addition to examining the existing models, the present comparison also provides several key considerations for testing the scientific soundness of any model of shear failure in concrete beams, which is necessary for future revisions to the design code provisions.