In a rabbit model, regeneration of the intratemporal facial nerve was studied as vascularized and nonvascularized nerve grafts. Fifteen New Zealand white rabbits weighing 4 to 4.5 kg were used. A 1-cm gap was created in the intratemporal facial nerve. In the vascularized nerve graft group (group 1), the facial nerve was repaired with a vascularized median nerve graft (n = 6), while in the nonvascularized nerve graft group (group 2), the median nerve was used as a nonvascularized graft (n = 6). In group 3, the facial nerve gap was left unrepaired (n = 3). At 3 months postoperatively, electrophysiologic testing, morphologic nerve study, and morphometric muscle analysis were performed. Although the mean numerical values of axonal counting, nerve conduction, and morphometric muscle study results were slightly better in the vascularized nerve graft group, the differences between the two grafted groups were not statistically significant (p > 0.05). Morphometric nerve analysis, i.e., axon diameter and myelin sheath thickness, showed significant differences in nerve regeneration between the two groups (p < 0.05). These results suggest that a bony bed is not optimal for neovascularization of a nerve graft. We conclude that vascularized nerve grafts are superior to nonvascularized nerve grafts when nerve regeneration occurs in a bony recipient bed. This might be of clinical importance in the repair of facial nerve lesions within the temporal bone.