Forty-three patients with peripheral neuropathic pain, exclusively pain reduced by spinal cord stimulation (SCS), were switched into a painful state after SCS inactivation. This mode was used to assess the pain-relieving effect of carbamazepine (CMZ) and opioids in a double-blinded, placebo-controlled trial. In Phase 1, the patients were randomly allocated to receive either CMZ (600 mg/d) or placebo during an SCS-free period of 8 days. In Phase 2, after a CMZ elimination interval of 7 days, 38 patients received either sustained-release morphine (90 mg/d) or placebo for 8 days. In cases of intolerable pain, the patients were authorized to reactivate their SCS. The pain intensity was rated on a numeric analog scale. In 38 patients who completed Phase 1, significant delay in pain increase was observed in the CMZ group as compared with placebo (P = 0.038). In Phase 2, the trend observed with morphine was insignificant (P = 0.41). Two CMZ patients and one morphine patient showed complete pain relief and preferred to continue the medication. Thirty-five patients returned to SCS. We conclude that CMZ is effective in peripheral neuropathic pain. Morphine obviously requires larger individually titrated dosages than those used in this study for results to be adequately interpreted.