Drugs that are clinically effective (mexiletine and desipramine) or ineffective (fluoxetine) in the treatment of human neuropathic pain were evaluated for efficacy in rat models involving central sensitization (i.e., formalin model and the L5/L6 spinal nerve ligation model of neuropathic pain) using tests that differ in stimulus modality: noxious chemical stimulus (formalin model) as well as noxious (pin prick) and innocuous mechanical stimuli (application of von Frey filaments). Mexiletine (10-100 mg/kg, s.c.) significantly (P < 0.05) attenuated hyperalgesia in formalin-treated (60 mg/kg and 100 mg/kg) and neuropathic rats (100 mg/kg) as well as tactile allodynia in neuropathic rats (100 mg/kg). Desipramine (1-100 mg/kg, s.c.), on the other hand, reduced hyperalgesia significantly (P < 0.05) in formalin-treated (3, 10, 30 and 100 mg/kg) and neuropathic rats (10 mg/kg and 100 mg/kg), but did not reduce tactile allodynia in the neuropathic rats. Fluoxetine (3-30 mg/kg, s.c.) did not inhibit either hyperalgesia or allodynia in any of the tests employed. Fluoxetine, which is relatively ineffective in reducing neuropathic pain in humans, was also ineffective in reducing hyperalgesia and allodynia associated with central sensitization in rats. Thus, drugs which are effective in reducing human neuropathic pain consistently attenuated hyperalgesia in formalin-treated or neuropathic rats. Desipramine also distinguished mechanical hyperalgesia from tactile allodynia in rats rendered neuropathic by spinal nerve ligation. These data are consistent with the hypothesis that the neuronal mechanisms underlying these two manifestations of neuropathic pain are different.