The present work defines a modified critical size rabbit ulna defect model for bone regeneration in which a non-resorbable barrier membrane was used to separate the radius from the ulna to create a valid model for evaluation of tissue-engineered periosteal substitutes. Eight rabbits divided into two groups were used. Critical defects (15 mm) were made in the ulna completely eliminating periosteum. For group I, defects were filled with a nanohydroxyapatite poly(ester urethane) scaffold soaked in PBS and left as such (group Ia) or wrapped with a tissue-engineered periosteal substitute (group Ib). For group II, an expanded-polytetrafluoroethylene (e-PTFE) (GORE-TEX(®)) membrane was inserted around the radius then the defects received either scaffold alone (group IIa) or scaffold wrapped with periosteal substitute (group IIb). Animals were euthanized after 12-16 weeks, and bone regeneration was evaluated by radiography, computed microtomography (μCT), and histology. In the first group, we observed formation of radio-ulnar synostosis irrespective of the treatment. This was completely eliminated upon placement of the e-PTFE (GORE-TEX(®)) membrane in the second group of animals. In conclusion, modification of the model using a non-resorbable e-PTFE membrane to isolate the ulna from the radius was a valuable addition allowing for objective evaluation of the tissue-engineered periosteal substitute.