Aim: To elucidate the influence of initial soft tissue thickness on peri‐implant bone remodeling. The research hypothesis was that implants installed in patients or at sites with thin mucosal tissues would show increased peri‐implant bone loss.

Material and Methods: 79 edentulous patients were consecutively treated with two non‐splinted implants supporting an overdenture in the mandible. During recall‐visits, peri‐implant health was determined by means of probing pocket depth and the modified plaque/bleeding index. Digital peri‐apical radiographs were taken from individual implants. Bone level changes were measured from a reference point (lower border of the smooth implant collar) to the marginal bone‐to‐implant contact level. The linear mixed‐effect model analysis was adopted to analyze the influence of clinical parameters and transmucosal abutment height on peri‐implant bone loss.

Results: 67 patients attended the 1‐year and 66 the 2‐year recall‐visit. Mean bone level changes were 0.89 mm ( SD 0.62) and 0.90 mm ( SD 0.66), plaque scores 0.82 ( SD 0.94) and 0.87 ( SD 0.92), bleeding scores 0.46 ( SD 0.68) and 0.56 ( SD 0.72) and PPD 1.65 mm ( SD 0.60) and 1.78 mm ( SD 0.59) after 1 year and 2 years respectively. The linear mixed‐effect model revealed increasing bone level changes with decreasing abutment heights. Peri‐implant bone level changes were significantly higher for implants with abutments of <2 mm (1.17 mm, p  < .01; 1.23 mm, p  < .01), 2 mm (0.86 mm, p  < .01; 1.03 mm, p  < .01) or 3 mm (0.38 mm, p  = .046; 0.41 mm, p  = .044) compared to ≥4 mm‐abutments (bone level changes set to zero as reference value) both after 1 year and 2 years and bone level changes were significantly influenced by probing pocket depth ( p  < .01, p  < .01), but not by plaque ( p  = .31, p  = .09) and bleeding scores ( p  = .30, p  = .40).

Conclusion: The present study suggests that implants with lower abutments, reflecting the initial gingival thickness, lose more peri‐implant bone, possibly by a re‐establishment of the biological width.