The purpose of this study was to compare bone regeneration within a surgically created defect using three different resorbable membranes (polyglactin 910 knitted mesh/Vicryl, freeze-dried fascia lata, and crosslinked bovine type I collagen/BioMend) and two nonresorbable membranes (expanded polytetrafluoroethylene/Gore Tex, and polytetrafluoroethylene/Millipore). Each of three adult male dogs underwent surgical preparation of six bicortical defects of the calvarium, each 1.5 cm in diameter. The experimental barrier membranes were placed on both sides of the defects (inner and outer cranial tables). Five of the six defects were covered randomly by one of the five membrane materials, and one was left uncovered to serve as a negative control. The animals were sacrificed 10 weeks after membrane placement. Bone response was measured clinically, as well as radiographically, via densitometric examination. All surgical sites healed uneventfully. Variable degrees of new bone growth were present at all sites when evaluated by both clinical and radiographic examination. The general trend of observed osseous response indicates a greater, although not statistically significant, degree of bone growth using nonresorbable membranes. The animal model employed appears to be an efficient and reliable means of evaluating guided bone regeneration membranes. Schlagwörter: bone regeneration, canine calvarium, guided tissue regeneration, resorbable membranes