From the study, the RTVs of the control group, which was left unloaded for 10 min after second tightening decreased 7.78% from the tightening torque. Previous studies have shown a 2 to 10% loss of screw preload [21]. These results can be explained by the settling effect as follows [16]. Firstly, the tightening torque is used to overcome the friction of the contacting metal surface connection [32]. Secondly, wear of the contacting implant-abutment surface can cause axial displacement of the abutment into the implant bore and the length of the elongated screw is shortened microscopically, leading to the loss of screw preload [12, 33]. After mechanical cyclic loading, all experimental groups from 50,000 to 2,000,000 cycles showed significant decreases in mean RTVs compared with the initial RTVs (P < 0.05). These results are similar to those of several studies. Mohammed et al. [17] reported that the post-loading RTVs of internal hex implants were significantly lower than the initial RTVs after 16,000 cycles and Cibirka et al. [26] found decreasing RTVs in the internal hexagon of Nobel Biocare implants after 5 million cycles of fatigue testing. The decrease in RTVs after cyclic loading can be explained by the micromovement of the joint connection or progressive settling effect from functional loading [33]. It is assumed that mechanical cyclic loading will serve as a proxy for oral functional loading, which can cause micromovement and slipping between the abutment screw thread and the implant, reducing the tensional force and resulting in decreased preload of the screw [34, 35]. Additionally, in functionally loaded implants, the progressive settling effect and the wedge effect cause increased axial displacement of the abutment into the implant connection [36]. Seol et al. [37], who analyzed the axial displacement of an internal implant-abutment connection after cyclic loading, found that the two-piece abutment of an internal octagon connection showed continuous axial displacement, but the rate of axial displacement was slow after 100,000 cycles. In our study, the decrease in RTVs was also constant after 50,000 cycles. This might imply that the axial displacement has a great effect on the loss of screw preload. Additionally, Kim et al. [38] indicated that the RTVs of abutment screws are related to the settling value. They found that after cyclic loading, there were statistically significant differences in the settling value and also in RTVs in internal connections, whereas external connections showed no significant changes in settling values and resulted in no significant changes in RTVs. However, the relation between settling values and RTVs varies depending on many factors, such as the material, design, and characteristic of the abutment-implant interface [38].