In this review, we compared the outcome of 25 studies of experimentally induced pancreatitis in animals with 13 studies of human acute pancreatitis in which the same therapeutic agents were used (aprotinin, glucagon, 5-fluorouracil, somatostatin, peritoneal lavage). Whereas 81% of the animal studies had a positive outcome (improvement in survival), only 7.7% of the human studies showed a positive outcome on survival. Most animal studies suffered from a protocol in which treatment was not significantly delayed after induction of acute pancreatitis. Of the 12 human studies that showed no effect of treatment on survival, none had sufficient statistical power (1 - beta error) for the investigators to have confidence in the negative outcomes. This was due to the fact that the studies had too few patients or that the event rates in the untreated populations were too low. Only five of the human studies reported the complication rates of acute pancreatitis in patients who did not die of their disease. Treatment (by any agent) did not improve the complication rate in these studies, but only one of the five reports had sufficient statistical power for the investigators to have confidence in these negative results. Large multicenter studies with sufficient numbers of patients with severe pancreatitis (high mortality and complication rates) are needed to evaluate new therapies in this disease.