Adequate treatment of cancer pain remains a significant clinical problem. To reduce side effects of treatment, intrathecal and epidural routes of administration have been used where appropriate to reduce the total dose of agent administered while achieving regional control. Resiniferatoxin (RTX), an ultrapotent capsaicin analog, gives long-term desensitization of nociception via C-fiber sensory neurons. We evaluate here the analgesic effect on rats of epidurally administered RTX, using latency of response to a thermal stimulus in unrestrained animals. Results were compared with those for systemically administered RTX. Vehicle or graded doses of RTX were injected subcutaneously (s.c.) or through an indwelling lumbar (L4) epidural catheter as a single dose. Both routes of application of RTX produced profound thermal analgesia, reaching a plateau within 4-6 h and showing no restoration of pain sensitivity over 7 days. Vehicle was without effect. For the epidural route, the effect was selective as expected for the targeted spinal cord region, whereas the subcutaneous administration of RTX had a generalized analgesic effect. At doses yielding a tripling of back paw withdrawal latency, epidural treatment was 25-fold more effective than the subcutaneous route of application. Consistent with the regional selectivity of the lumbar epidural route, the front paws showed no more effect than by systemic RTX treatment. Binding experiments with [3H]RTX provided further evidence of the segmental desensitization induced by epidural RTX. We conclude that epidural administration of RTX at the lumbar spinal level produces profound, long-lasting, segmental analgesia to C-fiber mediated pain in the rat.