Objectives.To evaluate the efficacy and mechanism of action of a protease inhibitor (ulinastatin) in septic shock. Design.Prospective, randomized, controlled trial. Setting.A university laboratory. Subjects.Twelve mongrel dogs. Interventions.One of the protease inhibitors, ulinastatin, a glycoprotein (molecular weight 67,000 daltons) detected in human urine was estimated. We used Escherichia coli to obtain a model of septic shock in dogs in vivo study. Human neutrophils were used as an activating target in vitro. Measurements and Main Results.The final concentration ofE. coli was 1.9×106 colony-forming units/mL. There was no significant difference in E. coli concentration between ulinas-tatin-treated and control groups. Human neutrophils treated with 100 U/mL of ulinastatin showed 70.5% to 78.7% of the superoxide production or untreated neutrophils. Phagocytic activity was enhanced in a dose-dependently manner by ulinastatin. At a ulinastatin concentration of 100 U/mL, an approximate two-fold increase in activation was found. In the ulinastatin-treated group, cardiac index, blood pressure, lactic acid, blood glucose, and blood base values significantly improved 60 mins after ulinastatin administration, and the bacterial count was significantly decreased, while the endotoxin concentration in the control group showed a continuous increase of endotoxin concentration. The improvement in the monitored factors observed 60 mins after initiation of treatment persisted after the end of treatment. The survival rate after 1 wk in the ulinastatin-treated group was 84% (five of six dogs survived), while it was 16% (one of six dogs survived) in the control group (p = .04). Conclusions.Ulinastatin does not have antimicrobial activity, and it does not sufficiently activate phagocytes. It is suggested that the efficacy of this agent in experimental septic shock is due to a mechanism that activates the reticuloendothelial system and septic reactions. (Crit Care Med 1993; 21:925–930)