Tensile bond strength data for resin-bonded bridge cements would predict failure loads in excess of 1000 N when based on available bonded surface area. Such loads are unlikely to be encountered clinically, yet these bridges frequently fail. Loading conditions resulting in a peeling action at the adhesive interface may provide a more probable explanation for failure rather than being attributable to poor clinical or laboratory technique. This hypothesis was investigated by bonding grit-blasted Ni/Cr alloy beams of different thicknesses to a block of the same alloy using a resin-bonded bridge cement, leaving half the length of the beam free. Beams were pulled off the block with a peeling action by applying a tensile load to the free end of the beam and load at failure recorded. Fractured surfaces were examined under SEM. A two-dimensional finite element model of a section through the peel test arrangement was used to produce data of the stress distribution in the adhesive layer. Load at failure was closely correlated to beam thickness. Using this data and known tensile bond strength values it was possible to predict the loads necessary to cause adhesive failure of the beam, which were found to correlate closely with the experimental measurements. It was concluded that design and the stiffness of the retainers can have a potentially profound effect on the adhesive quality of resin-bonded bridges.