The analgesic efficacy of tramadol and its M1 metabolite has been established in various rodent models of pain (Hennies et al 1988 ; Kayser et al 1991 ; Raffa et al 1992 , 1995 ; Mattia et al 1993 ). Competitive binding to rodent brain membranes in vitro showed that tramadol is a selective μ-opioid receptor agonist with a 200-fold lower affinity for μ-opioid receptors than its M1 metabolite (Hennies et al 1988 ; Raffa et al 1995 ). Opioid agonist activity appeared to be mediated by the dextrorotatory (+) enantiomeric form of tramadol, as opposed to the mirror image levorotatory (–) form (Raffa et al 1995 ). Tramadol is a relatively weak μ-opioid agonist; its affinity for μ-opioid receptors was 6000, 60, and 10 tmes lower than that of morphine, dextropropoxyphene, and codeine, respectively (Raffa et al 1995 ). Tramadol’s μ-opioid receptor binding and analgesic effects were only partially blocked by naloxone, suggesting other nonopioid mechanisms of analgesia (Hennies et al 1988 ; Raffa et al 1992 ). In vitro studies, using rodent synaptosomes, showed that tramadol was more potent with respect to norepinephrine and serotonin reuptake inhibition than μ-opioid receptor binding (Hennies et al 1982 ; Raffa et al 1992 ). That tramadol mediated its analgesic effects through inhibition of monoamine reuptake in vivo was confirmed in animal and human studies, in which tramadol’s analgesic effects were blocked by yohimbine (Raffa et al 1992 ; Desmeules et al 1996 ). Intrathecal tramadol produced a weak analgesic effect in mice (Mattia et al 1993 ). In contrast, administration of tramadol by the intracerebroventricular route followed by intrathecal injection resulted in powerful synergistic analgesia, implicating both the brain and spinal cord as the principal sites of action (Mattia et al 1993 ). Unlike other centrally acting analgesics, tramadol demonstrated limited potential for tolerance in mouse and rat pain models (Kayser et al 1991 ; Mattia et al 1993 ).