Introduction/Objective: To review the evidence on the hypothesis that placebo analgesia in humans is mediated by endogenous opioids.

Methods: A systematic review of experimental studies identified through computer assisted searching in MEDLINE (1966-1996), EMBASE (1984-1996) and PsycLIT (1974-1996), checking references, forward searching through the Science Citation Index, and personal communication with experts. A list of methodological criteria was used to assess systematically the strengths and weaknesses of eligible studies.

Results: Seven studies were appraised. Three studies provide evidence that a true placebo analgesic effect exists, which cannot be explained solely by response bias, that is, by assuming that subjects treated with placebo tend to change their pain ratings in accordance with their perception of investigators' wishes whereas no-treatment controls may feel no need to similarly change their pain ratings having not received (placebo) treatment. In the three studies in which response bias was excluded there is some evidence that endogenous opioids are involved in the placebo analgesic response. However, there is no consensus as to the requirements for truly hidden infusion and this hampers a consistent interpretation of the findings from the different studies. There is a clear need for replication of findings.

Discussion: There is some, as yet inconclusive, evidence that endogenous opioids play a role in mediating placebo analgesia.