Objective: This retrospective observational pilot study examined differences in peri-implant bone level changes (ΔIBL) between two similar implant types differing only in the surface texture of the neck. The hypothesis tested was that ΔIBL would be greater with machined-neck implants than with groovedneck implants. Method and Materials: 40 patients were enrolled; n = 20 implants with machined (group 1) and n = 20 implants with a rough, grooved neck (group 2), all placed in the posterior mandible. Radiographs were obtained after loading (at 3 to 9 months) and at 12 to 18 months after implant insertion. Case number calculation with respect to ΔIBL was conducted. Groups were compared using a Brunner-Langer model, the Mann-Whitney test, the Wilcoxon signed rank test, and linear model analysis. Results: After the 12- to 18-month observation period, mean ΔIBL was −1.11 ± 0.92 mm in group 1 and −1.25 ± 1.23 mm in group 2. ΔIBL depended significantly on time (P .001), but not on group. In both groups, mean marginal ΔIBL was significantly less than −1.5 mm. Only insertion depth had a significant influence on the amount of periimplant bone loss (P = .013). Case number estimate testing for a difference between group 1 and 2 with a power of 90% revealed a sample size per group of 1,032 subjects. Conclusion: ΔIBL values indicated that both implant designs fulfilled implant success criteria, and the modification of implant neck texture had no significant influence on ΔIBL. Palabras clave: implant collar, implant collar design, microgrooves