Vasopressin is currently recommended in the management of patients with cardiac arrest, but its efficacy is still incompletely established. We systematically reviewed randomized trials comparing vasopressin to control treatment in the management of cardiac arrest in humans and animals. Two human and 33 animal studies were retrieved. At pooled analysis vasopressin appeared equivalent to adrenaline (epinephrine) in the management of human cardiac arrest (N=240), with, respectively 63 (78/124) vs 59% (68/116) ROSC (P=0.43), and 16 (20/124) vs 14% (16/116) survival to hospital discharge (P=0.52). In animal trials (N=669) vasopressin appeared instead significantly superior to both placebo (ROSC, respectively 93 [98/105] vs 19% [14/72], P<0.001) or adrenaline (ROSC, respectively 84 [225/268] vs 52% [117/224], P<0.001). In conclusion, vasopressin is superior to both placebo or adrenaline in animal models of cardiopulmonary resuscitation. Evidence in humans is still limited and confidence intervals estimates too wide to reliably confirm or disprove results obtained in experimental animal settings.