An experimental model is presented for the local administration of neurotrophic substances at the site of peripheral nerve lesion. The model consists of a subcutaneously implanted silicone reservoir and a connecting tube with its distal end facing the severed and repaired nerve. Wistar rats (n = 180) were divided into two groups: a control group (saline-treated) (n = 90) and an NGF-treated group (n = 90). After sciatic nerve axotomy, an epineural repair was performed. NGF or saline were injected daily into the subcutaneous reservoir for the first 4 weeks after axotomy and weekly single dose between the 8th and 12th weeks. Both groups were divided into three subgroups of 30 animals each. The animals were sacrificed at 4, 8 and 12 weeks. Myelinated and non-myelinated axonal and thickness of myelin sheaths were quantified at the tibialis branch 25 mm distal to the nerve repair site. Axonal counts showed statistically significant differences between the treated and control groups at 4, 8 and 12 weeks. Finally, at 4 weeks the myelinated axons in the NGF group had significantly thicker myelin sheaths than in the control group. In comparison with other models of administration of different neurotrophic agents, NGF delivered through this system demonstrates a significant capacity for improving nerve regeneration without the problems inherent in multiple anesthesia, device exchange, or short half-life of the NGF single-dose administration.