Two full-scale, precast, pretensioned box girders were subjected to shear-dominated loading, one under monotonic loads to failure and the other subjected to one-half million cycles of fatigue loads followed by monotonic ultimate loads. The number of cycles was selected to allow for comparison with previous research. The fatigue loads were applied in combination with occasional overloads. In the present study, fatigue loading reduced the shear capacity by only six percent compared to the capacity under monotonic loading. However, previous research on flexure-dominated girders subjected to the same number of repeated loads showed that fatigue loading changed the mode of failure from flexure to shear/flexure and the girder capacity dropped by 14 percent. The comparison of the measured data with calculated shear capacity from five different theoretical methods showed that the ACI code method, the compression field theory, and the modified compression field theory led to reasonable estimates of the shear strength. The truss model led to an overly conservative estimate of the capacity.