Because severe sepsis is frequently complicated by multiple organ failure, it is of importance to monitor organ function. Unfortunately, conventional liver function markers are either relatively unspecific or have a long half-life, which make them poor predictors of acute liver injury. Glutathione S-transferase A1-1 (GSTA1-1) has a relatively short half-life (1 h), is more specific, and is rapidly released into the blood after liver damage. In the present study, we measured plasma GSTA1-1 levels by enzyme-linked immunosorbent assay in seven healthy volunteers after repeated experimental endotoxemia induced by 2 ng kg Escherichia coli endotoxin per day (to investigate inflammation-induced hepatic injury) and in 21 patients within 12 h after the occurrence of severe sepsis/septic shock (to investigate its ability to predict an increase of transaminases on day 7). During repeated experimental endotoxemia in healthy volunteers, TNF-alpha and IL-6 levels increased from undetectable levels to 1,425 (474-1,949) and 1,739 (989-2,047) pg mL, respectively, whereas GSTA1-1 levels did not exceed the normal range, indicating that no (sub)clinical liver injury occurs in this model of inflammation. In septic patients, GSTA1-1 levels had a specificity of 88%, resulting in a positive predictive value for liver injury of 86% and a positive likelihood ratio of 6 to indicate an increase in transaminases on day 7. Furthermore, GSTA1-1 levels did not correlate with IL-6 levels but did with dobutamine infusion rate (Spearman r = 0.94; P = 0.02), suggesting that the extent of hemodynamic instability and not the degree of inflammation could be of importance for the occurrence of liver damage. In septic shock patients, GSTA1-1 may represent a useful marker for early liver injury.