This study evaluated the effect of adding serration to the abutment-implant connection on torque maintenance before and after loading. Two implant systems with the same dimensions and connection design (internal 8˚morse taper octagon) were selected: one with non-serrated abutments (Simple line II) and the other one with serrated abutments (F & B). The removal torque value (RTV) was measured in two groups for each system: one group with one-piece abutments and the other group with two-piece abutments, before and after cyclic loading (n=10 in each group). The initial RTV of the abutment screw was measured with a digital torque meter. Each abutment received a cement-retained metal crown with 30° occlusal surface. Cyclic axial peak load of 75±5 N was applied to the implants for 500,000 cycles at 1 Hz. The post-load RTV was then measured. Two-way and repeated measures ANOVA, and independent t-test were applied to assess the effects of cyclic loading, connection design, abutment type, and their interaction on the percentage of torque loss (α=.05). Two-way ANOVA showed that serration of mating surfaces had a significant effect on torque maintenance before (P<0.001) and after (P=0.004) cyclic loading. Repeated measures ANOVA also showed that loading had a significant effect on the torque loss percentage (P< 0.01). Comparison of the groups with t-test showed that the torque loss of the serrated groups was lower than that of non- serrated groups. Despite the limitations of this study, the stability of the implant-abutment connection in the serrated design was higher than that of non-serrated group.