OBJECTIVES: Tumor necrosis factor (TNF) is an important mediator in the complex pathophysiology of sepsis syndrome. Although a positive correlation with mortality rate has been demonstrated, TNF has not been found consistently in sepsis. Since prolonged increases in soluble TNF receptor concentrations were demonstrated after endotoxin and TNF administration, we investigated whether the measurement of TNF receptor concentrations could provide a better indicator of disease than plasma TNF and interleukin (IL)-6 concentrations. DESIGN: Prospective analysis. SETTING: General intensive care unit (ICU) of a university hospital. PATIENTS: Twenty-six patients with sepsis syndrome and proven bacteremia. MEASUREMENTS AND MAIN RESULTS: Plasma peak concentrations of the soluble 55-kilodalton molecular weight TNF receptor were significantly higher (p < .005) in nonsurvivors compared with survivors of sepsis syndrome, whereas the difference in peak concentrations of the soluble 75-kilodalton TNF receptor did not reach significance (p = .06). In contrast to TNF peak concentrations (p = .14), significantly higher (p < .05) IL-6 peak concentrations were measured in nonsurvivors. Besides the positive correlation between the soluble 55-kilodalton TNF receptor and the soluble 75-kilodalton TNF receptor (r2 = .68; p < .0001), peak concentrations of both soluble 55-kilodalton TNF receptor and 75-kilodalton TNF receptor correlated significantly with plasma creatinine values, an indicator of renal function (r2 = .60; p < .0001 and r2 = .44; p < .001, respectively). Plasma creatinine concentrations were significantly higher in nonsurvivors (p < .001). CONCLUSIONS: In the population studied, plasma-soluble TNF receptor concentrations correlated with outcome as well as with plasma creatinine concentrations. The data presented suggest that increased plasma-soluble TNF receptor concentrations in patients with sepsis syndrome are merely the result of renal failure complicating sepsis, and are similarly correlated with mortality rate.