The aim of this study was to investigate the neurotransmissions involved in the antinociceptive effect of tramadol in the formalin test, which is an animal model of acute and tonic pain. A subcutaneous injection of formalin produces a biphasic nociceptive response: phase 1 (0-10 min-acute pain) and phase 2 (21-60 min-tonic pain). Nociceptive activity is reduced greatly during the 10 min between these two phases. We measured in mice the effects of (+/-)-tramadol, and of (+)- and (-)-tramadol administered before the induction of pain by formalin, in the presence and absence of drugs that act on the opioidergic, serotonergic and noradrenergic systems (naloxone, ketanserin, fluoxetine, maprotiline). With respect to animals treated with formalin alone, (+/-)-tramadol and its enantiomers significantly reduced the duration of nociceptive behaviours (lifting, licking, favouring, shaking, and flinching of the formalin-treated paw) during phase 2. This effect was prevented by the 5-HT(2) receptor antagonist ketanserin, but not by naloxone which, on the contrary, was able to prevent the antinociceptive effect of morphine. Naloxone and ketanserin did not affect the duration of nociceptive behaviour in animals not treated with tramadol. Fluoxetine (a selective 5-hydroxytryptamine (5-HT) reuptake inhibitor), but not maprotiline (a selective norepinephrine reuptake inhibitor), potentiated the antinociceptive effect of (+/-)-tramadol. In conclusion, we demonstrate that the serotonergic pathway is responsible for the antinociceptive effect of tramadol in phase 2 of the formalin test, and that this effect is mediated by 5-HT(2) receptors.