Kallistatin, an endogenous serine proteinase inhibitor, is protective against sepsis in animal models. The aim of this study was to determine the plasma concentration of kallistatin in intensive care unit (ICU) patients with severe sepsis and septic shock and to determine their potential correlation with disease severity and outcomes. We enrolled 86 ICU patients with severe sepsis and septic shock. Their plasma concentrations of kallistatin, kallikrein, tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-6, and IL-8 were measured by enzyme-linked immunosorbent assay. The association of kallistatin levels with disease severity and patient outcomes was evaluated. The relationship between kallistatin and other biomarkers was also analyzed. Plasma kallistatin levels on day 1 of ICU admission were lower in patients with septic shock compared with patients with severe sepsis (p = 0.004). Twenty-nine patients who died in the hospital had significantly lower day 1 kallistatin levels than patients who survived (p = 0.031). Using the optimal cutoff value (4 μg/ml) of day 1 plasma kallistatin determined by receiver operating characteristic curves for 60-day mortality, we found that high kallistatin levels were associated with a preferable 60-day survival (p = 0.012) by Kaplan-Meier analysis and lower Sequential Organ Failure Assessment (SOFA) scores over the first 5 days in the ICU (p = 0.001). High kallistatin levels were also independently associated with a decreased risk of septic shock, the development of acute respiratory distress syndrome, and positive blood cultures. In addition, there were inverse correlations between day 1 kallistatin levels and the levels of TNF-α, IL-1β, IL-6, and C-reactive protein, and SOFA scores on day 1. Our results indicate that during severe sepsis and septic shock, a decrease in plasma concentrations of kallistatin reflects increased severity and poorer outcome of disease.