The purpose of this study was to compare spontaneous bone regeneration, osteoconduction, and bone autografting in critical size calvarial and mandibular defects (defects which do not heal spontaneously during the lifetime of the animal) that were protected from soft-tissue interposition. Eighteen adult mongrel dogs underwent osteotomies to create a unilateral 30-mm segmental defect in the midbody of the edentulated right mandible and bilateral 15-mm x 20-mm full-thickness window defects in the parietal bones. The defects were either left empty, implanted with coralline hydroxyapatite (HA) blocks, or autografted with iliac cancellous bone. All defects were protected with a macroporous titanium mesh and the segmental mandibular defects were additionally stabilized by internal plate fixation. Specimens were retrieved after 2 and 4 months and three undecalcified longitudinal central sections including the osteotomy interfaces were prepared from each specimen for histometry and histology. Sections were analyzed for volume fractions of bone, soft tissue, and implant using scanning electron microscopy, backscatter electron imaging and histometric computer software. In the mandibular model, the empty defects exhibited the greatest amount of bone formation after 4 months (47.3 percent), which was greater than the amount of bone in the autografted group (34.8 percent) and significantly greater than the amount of bone within the hydroxyapatite implants (19.0 percent, p < 0.05). In the cranial defects, the autografted specimens demonstrated the greatest volume fraction of bone after 4 months (27.3 percent), which was significantly greater than within both the empty defects (18.2 percent, p < 0.05) and the hydroxyapatite implants (18.2 percent, p < 0.05). New bone formation in the mandibular defects united the cut ends at 4 months regardless of treatment and originated predominantly from the periosteum which remained present only along the alveolar border after surgical closure. In the calvarial defects, periosteum was not preserved and bone regenerated centripetally, originating from the diploë without any evidence of dural osteogenesis. Bone bridging was incomplete in the empty cranial defects at 4 months. In both the mandibular and cranial specimens, new bone at 2 months was a mixture of woven and parallel fibered bone. At 4 months, the new bone had remodeled almost entirely into mature Haversian bone. This study demonstrated a remarkable ability of defect protection with a macroporous protective sheet to facilitate bone regeneration in critical size mandibular and cranial bone defects. When active osteogenic periosteum was present, as in our mandibular model, we concluded that defect protection alone was sufficient to allow for healing even of critical size defects. When periosteum was absent as in our cranial defects, the limited spontaneous bone formation benefited from the added contributions of cancellous grafting and osteoconductive implants, both of which promoted bone bridging across the defects. We suggest that in the future a resorbable macroporous protective sheet would be advantageous in comparison to a titanium mesh to facilitate bone regeneration by preventing soft-tissue prolapse and allowing the migration of mesenchymal cells and the proliferation of blood vessels from the adjacent soft tissues into the bone defect. Finally, this study identified the need to differentiate critical size defects into those with and without defect protection and periosteum.