The alleviation of neuropathic pain cannot be satisfactorily achieved by treatment with opioids. There is much evidence to indicate that the active site of morphine for inducing effective analgesia is in the raphe magnus nucleus, where serotonin (5-HT, 5-hydroxytryptamine) acts as a primary transmitter. Therefore, we developed the hypothesis that 5-HT released in the raphe magnus nucleus could be related to the effectiveness of morphine in two mice models of neuropathic pain, diabetic (DM)-induced neuropathy and sciatic nerve ligation (SL). Two weeks after a single administration of streptozotocin, or 10 days after sciatic nerve ligation, mice were subcutaneously (s.c.) injected with morphine at 3, 5 and 10 mg/kg. The antinociceptive effect of morphine was estimated in the tail-pinch test; 5-HT content was measured after induction of neuropathic pain by microdialysis followed by high-performance liquid chromatography with electrochemical detection (HPLC-ECD). Morphine produced as insufficient antinociceptive effect in SL mice at all doses compared with that in sham-operated mice, while in DM mice, morphine given s.c. at 5 and 10 mg/kg produced antinociceptive effects compared with those in non-diabetic mice, but not at 3 mg/kg. The 5-HT content of dialysates, expressed as AUC for 75 min, in SL and DM mice was less than that in control mice. However, morphine given s.c. at 5 mg/kg did not significantly affect 5-HT levels in both mice models compared to their controls. These results suggest that the decrease in 5-HT levels in the raphe magnus nucleus may be related to attenuation of the analgesic effect of morphine caused by the abnormal pain state found in diabetes and partial peripheral nerve injury.