Purpose: To investigate the insertion/pull-out performance of splints produced by hand casting, thermoforming, milling, and 3D printing. Materials and Methods: A total of 120 identical mandibular splints (n = 8 specimens per group) were manufactured with hand casting, thermoforming, milling, and 3D printing. The splints were stored in water at 37ºC for 10 days and then placed onto cobalt-chromium arches and fixed on one side. Forces were applied to the other side (centric, perpendicular 50 N, 1 Hz) at two different positions (teeth 46 and 44/45) to pull out, and the test was then reset. The number of pull-out cycles until failure was recorded. The fracture behavior of the splints was investigated and characterized as fracture in the loading position, fracture at the fixation, or combined fracture. Splints were pulled off until fracture as a control (v = 1 mm/minute). Finite element analysis was used to verify the results. Statistical analyses were conducted with one-way ANOVA, post hoc Bonferroni, Pearson correlation, and Kaplan-Meier log-rank tests (α = .05). Results: The mean pull-off cycles varied from 7,839 (V-Print) to 1,600,000 (Optimill) at the tooth 46 position (FDI numbering system) and from 9,064 (Splint Comfort) to 797,750 (Optimill) at the 44/45 position. Log-rank test showed significantly (P < .001) different pull-out cycles between the systems (chi-square: 61,792 to 122,377). The thickness of the splints varied between 1.6 ± 0.2 mm (Splint Comfort) and 2.3 ± 0.1 mm (V-Print). Thickness and number of cycles were correlated (Pearson: 0.164; P = .074). The pull-off forces of the control varied significantly (P ≤ .040), ranging from 13.0 N (Keysplint) to 82.2 N (Optimill) at the tooth 46 position and from 25.2 N (Keysplint) to 139.0 N (Optimill) at the 44/45 position. Conclusions: The milled and cast splints survived more pull-off cycles than the printed or thermoformed splints. The pullout performance showed differences among the tested splint systems and indicated the influence of the material properties and processing.