The timing and mechanisms of peripheral nerve revascularization were investigated using a 2-cm sciatic nerve graft model in 58 rats. Epineurial perfusion was consistently established by 48 hours and endoneurial perfusion by 72 hours. The pattern of endoneurial profusion was 'all-or-none' - either all or none of the vessels in a fascicle exhibited blood flow. Conventional allografts exhibited similar revascularization dynamics and patterns. Capping the ends of the autograft with Silastic significantly delayed revascularization; no flow was observed at 4 days, and only a peripheral rim of perfused fascicular vessels was observed at 7 days. These patterns suggested that the primary method of revascularization in the conventional graft was longitudinal inosculation; no evidence of peripheral neovascularization or dependence on the graft bed as a source of revascularization was observed. The introduction of a major histocompatibility complex barrier between the grafted tissue and the recipient animal did not alter the timing or the mechanics of blood flow reestablishment.