A comparative study was undertaken to evaluate end-to-end versus peripheral nerve graft repair in cranial nerve reconstruction. In 14 rats, the oculomotor nerve was sharply transected in the cavernous sinus and repaired either by end-to-end coaptation (n = 7) or by interposition of a peripheral nerve graft (n = 7). The results were evaluated 16 weeks after surgery by light and transmission electron microsurgery and by morphometric analysis. The degree of neuroma formation, fibrosis, and axonal disorganisation at the repair site was the same for both groups. Histologically, both end-to-end and graft repair groups revealed various degrees of axonal regeneration with myelinated nerve fibres in the distal nerve segments. In both groups, the number of nerve fibres distal to the repair site was increased compared to proximal to the repair (P < 0.001) but myelinated axon diameter was significantly less than that of control nerves (P < 0.001). No difference existed between the two repair groups in terms of mean myelinated axonal diameter. However, the number and density of myelinated axons was statistically greater in the graft group (P < 0.05). In conclusion, despite the disadvantage of two repair sites, peripheral nerve grafting results in equal or slightly superior axonal regeneration compared to an end-to-end repair in the rodent model of intracranial oculomotor nerve reconstruction. We speculate that this may be due to the structure of the peripheral nerve graft.