Lumbar puncture (LP) is an attractive route to deliver drugs to the nervous system because it is a safe bedside procedure. Its use for gene therapy has been complicated by poor vector performance and failure to target neurons. Here we report highly effective gene transfer to the primary sensory neurons of the dorsal root ganglia (DRGs) with self-complementary recombinant adeno-associated virus serotype 8 (sc-rAAV8) modeling an LP. Transgene expression was selective for these neurons outlining their cell bodies in the DRGs and their axons projecting into the spinal cord. Immunohistochemical studies demonstrated transduction of cells positive for the nociceptive neuron marker vanilloid receptor subtype 1, the small peptidergic neuron markers substance P and calcitonin generelated peptide, and the nonpeptidergic neuron marker griffonia simplicifolia isolectin B4. We tested the efficacy of the approach in a rat model of chronic neuropathic pain. A single administration of sc-rAAV8 expressing the analgesic gene prepro--endorphin (ppEP) led to significant (P < 0.0001) reversal of mechanical allodynia for >3 months. The antiallodynic effect could be reversed by the -opioid antagonist naloxone 4 months after gene transfer (P < 0.001). Testing of an alternative nonopioid analgesic gene, IL-10, alone or in combination with ppEP was equally effective (P < 0.0001). All aspects of the procedure, such as the use of an atraumatic injection technique, isotonic diluent, a low-infusion pressure, and a small injection volume, are consistent with clinical practice of intrathecal drug use. Therefore, gene transfer by LP may be suitable for developing gene therapy-based treatments for chronic pain.adeno-associated virus ͉ dorsal root ganglion ͉ gene therapy ͉ -endorphin ͉ IL-10