Eight groups of “Z” joint specimens were designed for conducting direct shear tests on steel keyed joints. The shape, size, number and joint type of steel keys were taken as experimental parameters, and the cracking load, ultimate load, residual load, and the relationship between vertical loading force and vertical relative displacement of specimens were investigated. The results reveal that the joints relied on the mechanical occlusion between the steel keys and concrete to transmit the shear forces, the load–displacement curve of the joints had a long development course, and the joints could bear large relative displacement. The bearing capacity of steel shear keyed joints increased with the increase in diameter and material strength of the tenon. The ultimate bearing capacity of the epoxied joints with 1 and 2 steel keys were respectively 151.39%, and 69.68% higher than dry joints. Meanwhile, the epoxied joints were found to become brittle and fail in more sudden manner. After direct shear failure, the shear resistance of epoxied joints was completely provided by the steel shear key. The elastic stiffness and bearing capacity of the epoxied joints was not controlled by the number of the steel shear keys, while the dry joints were sensitive with the number of keys. By adding 1 or 2 steel shear keys to a flat joint, the ultimate bearing capacity could be increased by 3.81 and 6.06 times for dry joints, and 43% and 33% for epoxied joints, respectively.