Objectives: To investigate the effect of implant-abutment connection and screw channel angle on screw stability by comparing a newly introduced and an established connection, before and after cyclic loading.

Materials and methods: Implants (N = 44) with Torcfit (TF) or Crossfit (CF) connection were divided to be restored with a straight (CFS and TFS) or an angled screw access channel (CFA and TFA) titanium-base abutment (n = 11). CFA and TFA received screw-retained crowns, whereas CFS and TFS received hybrid zirconia abutments and cement-retained crowns. The initial torque value (ITV) of each complex (ITV_I) and removal torque value (RTV) after 24 h (RTV_I) were measured. Screws were replaced with new ones, ITVs were recorded again (ITV_F), and crowns were cyclically loaded (2.4 million cycles, 98 N) to measure RTVs again (RTV_F). Percentage torque loss was calculated. Data were analyzed (α = 0.05).

Results: ITVs were similar among groups (p ≥ .089). CF led to higher RTVs (p ≤ .002), while CFS had higher RTV_Ithan CFA (p = .023). After 24 h, CFS had lower percentage torque loss than TF, while CFA had lower percentage torque loss than TFA (p ≤ .011). After cyclic loading, CF led to lower percentage torque (p < .001).

Conclusion: The implant-abutment connection affected the removal torque values. However, no screw loosening occurred during cyclic loading, which indicated a stable connection for all groups. Screw access channel angle did not affect screw stability after cyclic loading.