The aim of the present investigation was to study bone formation in an experimentally-produced furcation defect in the dog. 15 foxhound dogs (group A) and 4 large mongrel dogs (group B) were used. The 2nd and 4th mandibular premolars were extracted and the 3rd lower premolars (3P3) were assigned as experimental teeth. "Through and through" furcation defects, about 4 mm high and 3 mm wide, were first produced in the experimental teeth of the dogs in group A. Reconstructive surgery was subsequently performed in group A using a GTR technique. The dogs of group A were scheduled for biopsy 2 weeks (2 dogs), 1 month (2 dogs), 2 months (2 dogs), 4 months (3 dogs), 5 months (3 dogs) and 6 months (3 dogs) after GTR. The dogs in group B (4 animals) represented healthy, untreated pristine furcations and served as positive controls. Biopsies from the 3P3 regions were harvested, embedded in paraffin and prepared for histological analysis. Mesio-distal sections were cut with the microtome set at 7 microm. The sections were stained in hematoxylin and eosin, and Van Gieson. 3 sections, about 50 microm apart, and representing the central portion of the furcation site were selected for histological measurements. In group A, the proportions of various structures in the newly formed bone and marrow were assessed. In addition, the proportions of primary and secondary osteons, and the number of bone multicellular units (BMU)/mm2 mineralized bone tissue were determined. In the pristine furcations (group B), the histological analyses were performed in a corresponding area to that of the healing furcations. The results demonstrated that the process of bone formation in a large "suprabony" furcation defect can be divided into 3 different phases, namely, (i) the formation of a provisional connective tissue, (ii) the development of a primary bone spongiosa (including mainly woven bone), (iii) the replacement of the spongiosa by lamellar bone and bone marrow through processes of modeling and remodeling. The newly-formed trabeculae of woven bone were reinforced by the deposition of parallel-fibered bone and lamellar bone, a finding which substantiates the validity of the concept that woven bone possesses poor weight-bearing properties and, hence, needs to be re-inforced by a more mature type of bone. The modeling of the newly-formed bone resulted in the formation of (i) one large marrow space in the center of the furcation and, in addition, (ii) a smaller bone marrow space in the most coronal portion of the defect. At the end of the study (6 months), the bone marrow occupied a much larger space than in the bone tissue of pristine furcations. It was suggested that the process of modeling or remodeling of bone tissue in the furcation defect was not completed at the end of the study.