Besides reversing the underlying cause, the first line treatment for the symptoms of shock is usually the administration of intravenous fluids. If this method is not successful, vasopressors such as dopamine, dobutamine, adrenaline, noradrenaline and vasopressin are recommended. It is unclear if there is a vasopressor of choice, either for the treatment of particular forms of shock or for the treatment of shock in general. To assess the efficacy of vasopressors for circulatory shock in critically ill patients. Our main aim was to assess whether particular vasopressors reduce overall mortality. We also intended to identify whether the choice of vasopressor influences outcomes such as length-of-stay in the intensive care unit and health-related quality of life. We searched MEDLINE, the Cochrane Central Register of Controlled Trials, EMBASE, PASCAL BioMed, CINAHL, BIOSIS, and PsychINFO:all from inception to November 2003; for randomized controlled trials. We also asked experts in the field and searched meta-registries for ongoing trials. We included randomized controlled trials comparing various vasopressors, vasopressors with placebo or vasopressors with intravenous fluids for the treatment of any kind of circulatory failure (shock). Mortality was the main outcome. Two reviewers abstracted data independently. Disagreement between two reviewers was discussed and resolved with a third reviewer. We used random effects models for combining quantitative data. We identified eight randomized controlled trials. Reporting of methodological details was for many items not satisfactory: only two studies reported allocation concealment, and two that the outcome assessor was blind to the intervention. Two studies compared norepinephrine plus dobutamine with epinephrine alone in patients with septic shock (52 patients, relative risk of death 0.98, 95% confidence interval 0.57 to 1.67). Three studies compared norepinephrine with dopamine in patients with septic shock (62 patients, relative risk 0.88, 0.57 to 1.36). Two studies compared vasopressin with placebo in patients with septic shock (58 patients, relative risk 1.04, 0.06 to 19.33). One study compared terlipressin with norepinephrine in patients with refractory hypotension after general anaesthesia but there were no deaths (20 patients). The current available evidence is not suited to inform clinical practice. We were unable to determine whether a particular vasopressor is superior to other agents in the treatment of states of shock.