Direct bone-to-implant contact, defined as "osseointegration", is considered most optimal for long-term stability and survival of dental implants. However, the possibility of the formation of a tooth-like attachment apparatus around implants has also been demonstrated. The purpose of this study was to explore the formation of periodontal tissues around titanium implants using a novel and unique experimental model. After resection of the crowns of the maxillary canine teeth in nine mongrel dogs, the roots were hollowed to a depth of 5 mm leaving a thin dentinal wall. Slits were prepared in the cavity wall to create passages from the chamber to the periodontal ligament area. A custom-made, titanium implant was placed into the center of each chamber. Machined, titanium plasma sprayed (TPS) and sand blasted with large grit and acid attacked (SLA) surfaces were used. A collagen barrier was placed over the submerged chamber. Following 4 months of healing, jaw sections were processed for histology. Newly formed periodontal ligament, alveolar bone, and root cementum filled the space between the implant and the wall of the chamber. Ingrown bone was neither in contact with dentin nor with the implant. Thus, an interposed soft connective tissue layer was present. Healing by fibrous encapsulation was observed around most implants. However, cellular cementum was deposited on one TPS and one SLA implant and on the dentinal walls of the chamber. This study shows a remarkable capacity for new periodontal tissue formation at a site where no such tissues ever existed. Maintenance of original periodontal tissue domains most likely prevented osseointegration of the implants. The cementum layer deposited on two implants was likely formed through cementoconductivity rather than by differentiation of periodontal ligament cells upon contact with the implant surface.