There is pre-clinical evidence, involving several animal species, suggesting that opioid peptides play a role in the physiopathology of shock (endotoxic, hypovolemic, cardiogenic, spinal, anaphylactic). Many case reports have suggested that naloxone (an opiate antagonist) might be an effective treatment for shock in humans, but others have not supported such a point of view. This controversy led us to undertake a meta-analysis of the available evidence on the efficacy of naloxone as a treatment measure of shock in humans.To evaluate the effectiveness and safety of naloxone in human shock and to estimate the methodological quality of the clinical trials.Computerized bibliographic search up to December 2002, review of references of all papers found on the subject and contact with primary investigators of eligible studies.Randomized controlled trials evaluating naloxone in human shock, regardless of the patient's age (adult, child or neonate).Three independent reviewers extracted data on study design, intervention, outcome and methodological quality.Three independent readers reviewed 80 human publications and selected six clinical trials. Overall agreement on study selection was perfect (concordance: 100%). This meta-analysis includes six studies involving 126 patients with septic, cardiogenic, hemorrhagic or spinal shock. Naloxone therapy was associated with statistically significant hemodynamic improvement (odds ratio 0.24; 95% confidence interval [95%CI] 0.09-0.68). The mean arterial pressure was significantly higher in the naloxone groups than in the placebo groups (weighted mean difference: +9.33 mmHg; 95%CI 7.07-11.59). No heterogeneity was found for this outcome. The death rate was lower in the naloxone group (odds ratio 0.59; 95%CI 0.21-1.67) but this was consistent with the play of chance. A significant heterogeneity for the latter outcome was detected (p<0.05).Naloxone improves blood pressure, especially mean arterial blood pressure. However, the clinical usefulness of naloxone to treat shock remains to be determined, and additional randomized controlled trials are needed to assess its usefulness.