Different neurobiological mechanism(s) might contribute to evoked and non-evoked pains and to limited translational drug discovery efforts. Other variables including the pain model and sensory testing method used, dose/route/preadministration time of compound(s), lack of adverse effect profiling and level of observer experience might also contribute. With these points in mind, we tested three mechanistically distinct analgesics in rat models of algogen-induced and neuropathic pain. In chronic constriction injury (CCI) rats evoked hindpaw mechanical hypersensitivity and spontaneous weight bearing deficits developed quickly and persisted for at least 3 weeks post-injury. In contrast, evoked cold hypersensitivity, or movement-associated behavioural deficits (rotarod, beam-walking) were less manifested or dissipated rapidly post-injury. Mechanical hypersensitivity was dose-dependently reversed by morphine (3-10 mg/kg, s.c.) and gabapentin (50-200 mg/kg, i.p.). Weight bearing deficits and cold hypersensitivity were reversed only by high doses of each drug. Surprisingly, duloxetine (10-60 mg/kg, s.c.) was largely ineffective in neuropathic rats although it partially reduced formalin-induced spontaneous nocifensive behaviours; especially during interphase, a period associated with activation of descending monoaminergic inhibition. Morphine and gabapentin markedly attenuated second phase formalin- and in addition capsaicin-induced nocifensive behaviours; indicative of effects on central sensitization and nociceptor hyperexcitability mechanisms. Only gabapentin consistently attenuated nociceptive behaviours at a dose that did not impair exploratory locomotor behaviour in naïve rats. Accordingly, this comparative analysis indicates that the pharmacological sensitivity of evoked and non-evoked pain indices does not necessarily correlate within models, perhaps reflecting differing underlying mechanisms. Conversely, the pharmacological specificity of non-evoked pain indices to selected drugs was conserved across models indicative of similar underlying mechanisms enduring in the face of differing aetiology. Finally, although the predictive capacity of these models was largely unaffected by observer-related experience, it was putatively compromised when adverse event profiling of each drug was considered in parallel.