Obvious differences exist among the design provisions for interface shear transfer between concretes cast at different times (that is, cold joint) in current codes. The purpose of this study is to evaluate the applicability and accuracy of cold joint shear transfer provisions. Ten push-off tests on high-strength concrete with a cold joint were carried out in this paper, and the interface shear transfer mechanism was studied based on the tests. A database of 97 cold joint push-off test results of shear resistance, which was composed of the 10 test results from this paper and 87 test results collected from the literature, was created to evaluate the shear transfer provisions from the American Concrete Institute (ACI), AASHTO LRFD Bridge Design Specifications, the Precast/Prestressed Concrete Institute (PCI), and CSA-S6. The analysis results show that for the smooth interface condition, unconservative results may occur in using AASHTO in the case of low concrete strength; for the roughened interface condition, the experimental resistances are an average of 2.66 and 2.61 times greater than the nominal resistances from ACI and PCI, respectively, which indicates that ACI and PCI seem to be excessively conservative. Furthermore, whether for the smooth interface condition or for the rough interface condition, AASHTO and CSA-S6, which include the adhesion or cohesion term, have a higher accuracy and stability than ACI and PCI. It is in accordance with the analysis result of the interface shear transfer mechanism that adhesion or cohesion between two surfaces that form the interface significantly contribute to the shear resistance of the interface. In addition, the experimental-to-nominal shear resistance ratio k of all code provisions is positively correlated with the concrete strength, while no obvious relationship between k and the clamping stress of shear reinforcement is noted.