Reinforced concrete bridge deck slabs without shear reinforcement can be subjected to concentrated or distributed loads of important magnitude. Under these loads their structural response is not always ductile. In particular under concentrated loads their deformation capacity can be limited by shear or punching shear failures, which prevent them from reaching the ultimate load predicted by pure flexural analysis. This problem has been studied in this research by means of an important experimental program and theoretical modeling. The limited ductility of bridge decks was investigated by means of full scale tests on bridge deck cantilevers under groups of concentrated loads. Six large scale laboratory tests were performed on two bridge deck cantilevers with a span of 2.8 m and a length of 10.0 m. All slabs failed in a brittle manner, in shear or punching shear. The theoretical flexural failure load estimated using the yield-line method was never attained. Despite the brittle failures, the results of tests on cantilevers have shown that some amount of yielding can occur before the shear failure and therefore reduce the shear strength. This effect was quantified on eleven full scale tests on slab strips without shear reinforcement with a length of 8.4 m. The results clearly show that the increase of plastic strains in the flexural reinforcement leads to a reduction of the shear strength. The measured rotation capacity of the plastic hinge was thus limited by a shear failure. A particular problem of bridge deck slabs is the introduction of concentrated loads applied by wheels with pneumatic pressure. Punching shear with these loads is usually treated in a manner similar to punching by a column. A punching shear test was performed with a concentrated load simulating a vehicle wheel with pneumatic pressure to investigate the differences. It appears that punching shear with a wheel with pneumatic pressure is less critical because curvatures tend to be distributed over the surface of the applied load rather than concentrated near the edges of the column. In order to investigate the experimental results on slab strips without shear reinforcement, a mechanical model is proposed to predict the shear strength and rotation capacity of plastic hinges. The shear strength is formulated as a function of the opening of the shear crack and of the strength of the concrete compression zone. The results of the mechanical model are in good agreement with the measured values, both for the shear strength and for the shear carried across the shear crack. Based on the mechanical model, a simplified equation is proposed. The model can also be used to predict the shear capacity of yield-lines. A non linear finite element model was implemented during this work and used to correctly predict the measured rotations and load-displacements curves of the tested cantilevers and other full scale tests performed by other researchers. The measured failure loads are accurately estimated by using the results of the non-linear model and the one-way shear and punching shear criteria proposed by Prof. A. Muttoni (Muttoni 2003).