Pain is characterized by its multi-dimensional nature, explaining in part why the pharmacokinetic/pharmacodynamic (PK/PD) relationships are not straightforward for analgesics. The first part of this MiniReview gives an overview of PK, PD and PK/PD models, as well as of population approach used in analgesic studies. The second part updates the state-of-the-art in the PK/PD relationship of opioids, focusing on data obtained on experimental human pain models, a useful tool to characterize the PD of analgesics. For the so-called weak opioids such as codeine, experimental human studies showed that analgesia relies mainly upon biotransformation into morphine. However, the time-course of plasma concentrations of morphine did not always reflect the time-course of effects, the major site of action being the central nervous system. For tramadol, a correlation has been observed between the analgesic response and the PK of the (+)R-O-demethyl-tramadol metabolite. For ‘stronger’ opioids such as oxycodone, studies assessing the PK/PD of oxycodone suggested that active metabolite oxymorphone also strongly contributes to the analgesia and that analgesia may also be partially related through an action to peripherally located κ-opioid receptors. Different models have been proposed to describe the time-course of buprenorphine. An effect-compartment model was adopted to describe the PK/PD of morphine and its active metabolite, morphine-6-glucuronide (M6G). A longer blood-effect site equilibration half-life t1/2ke0 was observed for M6G, suggesting a longer onset of action. The studies assessing the PK/PD of fentanyl and its derivatives showed a short t1/2ke0 for analgesia, between 0.2 and 9 min., reflecting a short onset of effect. In conclusion, depending on the speed of transfer between the plasma and the effect site as well as the participation of active metabolites, the time-course of the analgesic effects can be close to the plasma concentrations (alfentanil and derivates) or observed with a prolonged delay (codeine, buprenorphine, morphine). These PK/PD data can be used to better characterize the differences between opioids, and partly explain the important observed variability among opioids in experimental conditions and should be systematically evaluated during drug development to better predict their selection in specific clinical conditions.