OBJECTIVE: To assess the diagnostic value of a single determination of serum C-reactive protein as a marker of sepsis in critically ill patients. DESIGN: Prospective, observational study. SETTING: Intensive care unit of a university hospital. PATIENTS AND PARTICIPANTS: One hundred twenty-five adult patients with systemic inflammatory response syndrome (SIRS) (55 patients without evidence of infection and 70 patients with the diagnosis of sepsis confirmed by documented infection). Twenty-five patients with non-complicated acute myocardial infarctions (AMI) and 50 healthy volunteers were used as controls. INTERVENTIONS: None. MEASUREMENTS AND RESULTS: Serum C-reactive protein concentration was measured within the first 24 h of SIRS onset. Healthy subjects, AMI and non-infectious SIRS patients showed lower C-reactive protein median values ([(0.21 [95% confidence intervals (95% CI), 0.21-0.4] mg/dl, 2.2 [95% CI, 2.1-4.9] mg/dl and 1.7 [95% CI, 2.4-5.5] mg/dl, respectively) than patients with sepsis (18.9 [95% CI, 17.1-21.8]), p<0.001. The presence of severe sepsis ( r(s)=0.27; p=0.03), SOFA score ( r(s)=0.25; p=0.03) and arterial lactate ( r(s)=0.24; p=0.04) correlated significantly with C-reactive protein concentrations in sepsis cases. The best threshold value for C-reactive protein for predicting sepsis was 8 mg/dl (sensitivity 94.3%, specificity 87.3%). The area under the receiver-operating characteristic curve for C-reactive protein was 0.94 (95% CI, 0.89-0.98). CONCLUSIONS: Determination of serum C-reactive protein can be used as an early indicator of infection in patients with SIRS.