Objective: This study investigated the effects of abutment screw withdrawal after conical abutment settlement on the stability of the implant–abutment connection. Materials and methods: Twenty implants of a conical connection system were used. Two two-piece abutment designs were used: cone only (n = 10; NI) and cone plus octagonal index design (n = 10; I); for each design, five samples were used with (S) and without (NS) abutment screw withdrawal before a cyclic test. Finally, four groups, namely Gr S(NI), Gr S(I), Gr NS(NI), and Gr NS(I), were included. The cyclic test included cyclic loading of 20–200 N, 30°, and 4-mm off-axis to implant axis at 10 Hz for 106 cycles, simulating a clinical time interval of 40 months. The fatigue cycles were recorded. The axial displacement of the conical abutments during abutment settlement, screw withdrawal, and cyclic loading were measured. Abutment morphology was examined through scanning electron microscopy (SEM). Results: Only Gr NS(NI) failed the test, indicating that without the index design and abutment screw withdrawal, and connection stability seriously deteriorated. Gr NS(I) exhibited significantly higher axial displacement into the implant after abutment settlement than did Gr NS(NI). It also exhibited continuous axial displacement into the implant after cyclic loading. SEM after cyclic testing in Gr NS(I) revealed marked burnishing on lateral edges of the index, indicating that the index design provides an antitorsional ability. Conclusion: Although this study has few limitations, abutment screw withdrawal is feasible in this conical implant–abutment connection system with index design.