Objective Targeted adenoviral gene delivery from peripheral nerves was used to integrally analyse the characterization and time course of LacZ gene (AdLacZ) retrograde transfer to spinal cord and transgene product anterograde labeling ofperipheral nerve. Methods Recombinant replication-defective adenovirus containing AdLacZ was administrated to the cut proximal stumps of median and tibial nerves in Wister rats. Then the transected nerve was repaired with 10-0 nylon sutures. At different time point postinfection the spinal cords of C5 to T1 attached with DRGs and brachial plexuses, or L2 to L6 attached with DRGs and lumbosacralplexuses were removed. The removed spinal cord and DRGs were cut into 50 μm serialcoronal sections and processed for X-gal staining and immunohistochemical staining. The whole specimens of brachial or lumbosacral plexuses attaching with theirperipheral nerves were processed for X-gal staining. The number of X-gal stained neurons was counted and the initial detected time of retrograde labeling, peaktime and persisting period of gene expression in DRG sensory neurons, spinal cord motor neurons and peripheral nerves were studied. Results The gene transfer was specifically targeted to the particular segments of spinal cord andDRGs, and transgene expression was strictly unilaterally corresponding to the infected nerves. Within the same nerve models, the initial detected time of gene expression was earliest in DRG neurons, then in the motor neurons and latest in peripheral nerves. The persisting duration of β-gal staining was shortest in motor neurons, then in sensory neurons and longest in peripheral nerves. The initial detected time of β-gal staining in median nerve models was earlier in mediannerve models compared with that in the tibial nerve models. Although the initial detected time and the beginning of peak duration of β-gal staining were not same, the decreasing time of β-gal staining in motor and sensory neurons of thetwo nerve models were started at about the same day 8 post-infection. The labeled neurons were more in tibial nerve-models than that in median nerve models. Within the same models, the labeled sensory neurons of DRGs were morethan labeled motor neurons of ventral horn. The β-gal staining was tenser in median nerves than that in tibial nerves. However the persisting time of β-gal staining was longer in tibial nerve models. Conclusion The b gene expression in neurons and PNS renders this system particularly attractive for neuroanatomical tracing studies. Furthermore this gene delivery method allowing specific targeting of motor and sensory neurons without damaging the spinal cord might offer potentialities for the gene therapy of peripheral nerve injury.