The present study evaluated the potential for reestablishment of connective tissue attachment after exclusion of the gingival connective tissue and epithelium during healing after periodontal surgery in a dog model. Buccal mucoperiosteal flaps were elevated in all four premolar regions in two dogs. The buccal bone was reduced to approximately 25% of its original level on 24 roots. Before the elevated flaps were replaced and sutured, membranes were adjusted to cover the exposed root surfaces. On eight surfaces Millipore filters were used and on eight surfaces biodegradable polylactic acid membranes were used. No membranes were placed over the remaining eight root surfaces which served as controls. After two months of healing, the animals were sacrificed and all the roots were subjected to histological analysis. In all of the roots covered with the polylactic acid membrane, newly formed cementum with inserting collagen fibers was observed and covered on the average 46% of the initial defect. Newly formed bone covered 39% of the initial defect. The roots covered by Millipore filters were characterized by significantly less new attachment and bone. In the control roots only minor new attachment was found in the bottom of the defect in some roots. The results suggest that the establishment of a connective tissue attachment is clearly favored by the placement of a polylactic acid membrane to exclude the epithelium and gingival connective tissue during healing.